活动介绍
file-type

OpenCV函数手册完整指南分享

版权申诉

ZIP文件

944KB | 更新于2024-10-23 | 125 浏览量 | 0 下载量 举报 收藏
download 限时特惠:#14.90
知识点: 1. OpenCV简介:OpenCV,全称Open Source Computer Vision Library,是一个开源的计算机视觉和机器学习软件库。它由一系列C函数和C++类构成,实现了图像处理和计算机视觉方面的很多常见算法。OpenCV具有跨平台的特性,在Windows,Linux,Mac OS,Android和iOS操作系统上都可以运行。 2. OpenCV功能:OpenCV的主要功能包括图像处理、视频分析、物体识别、图像分类、特征提取、机器学习等。它还提供了包括Python、Java等多种语言接口,方便开发者使用。 3. OpenCV函数:OpenCV库中包含了大量的函数,用于处理图像和视频数据。这些函数大致可以分为以下几类: - 基本图像处理函数:例如图像的读取、显示、保存、拷贝、裁剪、旋转、缩放等。 - 高级图像处理函数:例如图像滤波、图像变换、直方图操作、形态学操作等。 - 特征检测与描述函数:例如SIFT、SURF、ORB、BRISK等特征检测算法和描述算法。 - 对象检测函数:例如Haar级联分类器、HOG+SVM、深度学习目标检测等。 - 视频分析函数:例如背景减除、光流跟踪、运动分析等。 - 图像分割函数:例如阈值分割、分水岭算法、图割等。 - 机器学习函数:OpenCV中的机器学习模块提供了多种机器学习算法,包括支持向量机(SVM)、k-最近邻(kNN)、随机森林、梯度提升决策树等。 4. OpenCV学习资源:这份函数手册对OpenCV中的函数进行了详细说明,是学习OpenCV的重要资料。除了这份手册,还有许多在线资源可以帮助学习OpenCV,例如官方文档、教程、论坛、视频课程等。此外,还有一些相关的书籍,如《Learning OpenCV》、《OpenCV Computer Vision Application Programming Cookbook》等。 5. OpenCV应用:OpenCV可以应用于许多领域,包括机器人视觉、医疗成像、安全监控、交互式艺术、图像编辑、人机交互、移动应用、汽车驾驶辅助系统等。 6. OpenCV的安装和配置:安装OpenCV通常涉及到下载对应语言的库文件,然后在程序中导入使用。对于不同操作系统和开发环境,安装和配置的方式可能有所不同。 7. OpenCV的版本更新:OpenCV会不定期更新版本,每个新版本都会增加一些新的功能,改进一些现有的功能,修复一些bug。所以,定期更新OpenCV可以帮助开发者更好的利用库中的新功能。 以上就是这份OpenCV函数手册中可能包含的知识点。希望这些知识点能帮助大家更好的理解和使用OpenCV。

相关推荐

filetype

mport ctypes import os import shutil import random import sys import threading import time import cv2 import numpy as np import pycuda.autoinit import pycuda.driver as cuda import tensorrt as trt CONF_THRESH = 0.8 IOU_THRESHOLD = 0.6 def get_img_path_batches(batch_size, img_dir): ret = [] batch = [] for root, dirs, files in os.walk(img_dir): for name in files: if len(batch) == batch_size: ret.append(batch) batch = [] batch.append(os.path.join(root, name)) if len(batch) > 0: ret.append(batch) return ret def plot_one_box(x, img, color=None, label=None, line_thickness=None): """ description: Plots one bounding box on image img, this function comes from YoLov5 project. param: x: a box likes [x1,y1,x2,y2] img: a opencv image object color: color to draw rectangle, such as (0,255,0) label: str line_thickness: int return: no return """ tl = ( line_thickness or round(0.002 * (img.shape[0] + img.shape[1]) / 2) + 1 ) # line/font thickness color = color or [random.randint(0, 255) for _ in range(3)] c1, c2 = (int(x[0]), int(x[1])), (int(x[2]), int(x[3])) cv2.rectangle(img, c1, c2, color, thickness=tl, lineType=cv2.LINE_AA) if label: tf = max(tl - 1, 1) # font thickness t_size = cv2.getTextSize(label, 0, fontScale=tl / 3, thickness=tf)[0] c2 = c1[0] + t_size[0], c1[1] - t_size[1] - 3 cv2.rectangle(img, c1, c2, color, -1, cv2.LINE_AA) # filled cv2.putText( img, label, (c1[0], c1[1] - 2), 0, tl / 3, [225, 255, 255], thickness=tf, lineType=cv2.LINE_AA, ) class YoLov5TRT(object): """ description: A YOLOv5 class that warps TensorRT ops, preprocess and postprocess ops. """ def __init__(self, engine_file_path): # Create a Context on this device, self.ctx = cuda.Device(0).make_context() stream = cuda.Stream() TRT_LOGGER = trt.Logger(trt.Logger.INFO) runtime = trt.Runtime(TRT_LOGGER) # Deserialize the engine from file with open(engine_file_path, "rb") as f: engine = runtime.deserialize_cuda_engine(f.read()) context = engine.create_execution_context() host_inputs = [] cuda_inputs = [] host_outputs = [] cuda_outputs = [] bindings = [] for binding in engine: # print('bingding:', binding, engine.get_binding_shape(binding)) size = trt.volume(engine.get_binding_shape(binding)) * engine.max_batch_size dtype = trt.nptype(engine.get_binding_dtype(binding)) # Allocate host and device buffers host_mem = cuda.pagelocked_empty(size, dtype) cuda_mem = cuda.mem_alloc(host_mem.nbytes) # Append the device buffer to device bindings. bindings.append(int(cuda_mem)) # Append to the appropriate list. if engine.binding_is_input(binding): self.input_w = engine.get_binding_shape(binding)[-1] self.input_h = engine.get_binding_shape(binding)[-2] host_inputs.append(host_mem) cuda_inputs.append(cuda_mem) else: host_outputs.append(host_mem) cuda_outputs.append(cuda_mem) # Store self.stream = stream self.context = context self.engine = engine self.host_inputs = host_inputs self.cuda_inputs = cuda_inputs self.host_outputs = host_outputs self.cuda_outputs = cuda_outputs self.bindings = bindings self.batch_size = 1 def infer(self, raw_image_generator): # threading.Thread.__init__(self) # Make self the active context, pushing it on top of the context stack. self.ctx.push() # Restore stream = self.stream context = self.context engine = self.engine host_inputs = self.host_inputs cuda_inputs = self.cuda_inputs host_outputs = self.host_outputs cuda_outputs = self.cuda_outputs bindings = self.bindings # Do image preprocess batch_image_raw = [] batch_origin_h = [] batch_origin_w = [] batch_input_image = np.empty(shape=[self.batch_size, 3, self.input_h, self.input_w]) input_image, image_raw, origin_h, origin_w = self.preprocess_image(raw_image_generator) batch_image_raw.append(image_raw) batch_origin_h.append(origin_h) batch_origin_w.append(origin_w) np.copyto(batch_input_image[0], input_image) batch_input_image = np.ascontiguousarray(batch_input_image) # Copy input image to host buffer np.copyto(host_inputs[0], batch_input_image.ravel()) start = time.time() # Transfer input data to the GPU. cuda.memcpy_htod_async(cuda_inputs[0], host_inputs[0], stream) # Run inference. context.execute_async(batch_size=self.batch_size, bindings=bindings, stream_handle=stream.handle) # Transfer predictions back from the GPU. cuda.memcpy_dtoh_async(host_outputs[0], cuda_outputs[0], stream) # Synchronize the stream stream.synchronize() end = time.time() # Remove any context from the top of the context stack, deactivating it. self.ctx.pop() # Here we use the first row of output in that batch_size = 1 output = host_outputs[0] # Do postprocess for i in range(self.batch_size): result_boxes, result_scores, result_classid = self.post_process(output[i * 6001: (i + 1) * 6001], batch_origin_h[i], batch_origin_w[i]) return result_boxes, result_scores, result_classid, end - start # # Draw rectangles and labels on the original image # for j in range(len(result_boxes)): # box = result_boxes[j] # plot_one_box( # box, # batch_image_raw[i], # label="{}:{:.2f}".format( # categories[int(result_classid[j])], result_scores[j] # ), # ) def destroy(self): # Remove any context from the top of the context stack, deactivating it. self.ctx.pop() def get_raw_image(self, image_path_batch): """ description: Read an image from image path """ for img_path in image_path_batch: yield cv2.imread(img_path) def get_raw_image_zeros(self, image_path_batch=None): """ description: Ready data for warmup """ for _ in range(self.batch_size): yield np.zeros([self.input_h, self.input_w, 3], dtype=np.uint8) def preprocess_image(self, raw_bgr_image): """ description: Convert BGR image to RGB, resize and pad it to target size, normalize to [0,1], transform to NCHW format. param: input_image_path: str, image path return: image: the processed image image_raw: the original image h: original height w: original width """ image_raw = raw_bgr_image h, w, c = image_raw.shape image = cv2.cvtColor(image_raw, cv2.COLOR_BGR2RGB) # Calculate widht and height and paddings r_w = self.input_w / w r_h = self.input_h / h if r_h > r_w: tw = self.input_w th = int(r_w * h) tx1 = tx2 = 0 ty1 = int((self.input_h - th) / 2) ty2 = self.input_h - th - ty1 else: tw = int(r_h * w) th = self.input_h tx1 = int((self.input_w - tw) / 2) tx2 = self.input_w - tw - tx1 ty1 = ty2 = 0 # Resize the image with long side while maintaining ratio image = cv2.resize(image, (tw, th)) # Pad the short side with (128,128,128) image = cv2.copyMakeBorder( image, ty1, ty2, tx1, tx2, cv2.BORDER_CONSTANT, None, (128, 128, 128) ) image = image.astype(np.float32) # Normalize to [0,1] image /= 255.0 # HWC to CHW format: image = np.transpose(image, [2, 0, 1]) # CHW to NCHW format image = np.expand_dims(image, axis=0) # Convert the image to row-major order, also known as "C order": image = np.ascontiguousarray(image) return image, image_raw, h, w def xywh2xyxy(self, origin_h, origin_w, x): """ description: Convert nx4 boxes from [x, y, w, h] to [x1, y1, x2, y2] where xy1=top-left, xy2=bottom-right param: origin_h: height of original image origin_w: width of original image x: A boxes numpy, each row is a box [center_x, center_y, w, h] return: y: A boxes numpy, each row is a box [x1, y1, x2, y2] """ y = np.zeros_like(x) r_w = self.input_w / origin_w r_h = self.input_h / origin_h if r_h > r_w: y[:, 0] = x[:, 0] - x[:, 2] / 2 y[:, 2] = x[:, 0] + x[:, 2] / 2 y[:, 1] = x[:, 1] - x[:, 3] / 2 - (self.input_h - r_w * origin_h) / 2 y[:, 3] = x[:, 1] + x[:, 3] / 2 - (self.input_h - r_w * origin_h) / 2 y /= r_w else: y[:, 0] = x[:, 0] - x[:, 2] / 2 - (self.input_w - r_h * origin_w) / 2 y[:, 2] = x[:, 0] + x[:, 2] / 2 - (self.input_w - r_h * origin_w) / 2 y[:, 1] = x[:, 1] - x[:, 3] / 2 y[:, 3] = x[:, 1] + x[:, 3] / 2 y /= r_h return y def post_process(self, output, origin_h, origin_w): """ description: postprocess the prediction param: output: A numpy likes [num_boxes,cx,cy,w,h,conf,cls_id, cx,cy,w,h,conf,cls_id, ...] origin_h: height of original image origin_w: width of original image return: result_boxes: finally boxes, a boxes numpy, each row is a box [x1, y1, x2, y2] result_scores: finally scores, a numpy, each element is the score correspoing to box result_classid: finally classid, a numpy, each element is the classid correspoing to box """ # Get the num of boxes detected num = int(output[0]) # Reshape to a two dimentional ndarray pred = np.reshape(output[1:], (-1, 6))[:num, :] # Do nms boxes = self.non_max_suppression(pred, origin_h, origin_w, conf_thres=CONF_THRESH, nms_thres=IOU_THRESHOLD) result_boxes = boxes[:, :4] if len(boxes) else np.array([]) result_scores = boxes[:, 4] if len(boxes) else np.array([]) result_classid = boxes[:, 5] if len(boxes) else np.array([]) return result_boxes, result_scores, result_classid def bbox_iou(self, box1, box2, x1y1x2y2=True): """ description: compute the IoU of two bounding boxes param: box1: A box coordinate (can be (x1, y1, x2, y2) or (x, y, w, h)) box2: A box coordinate (can be (x1, y1, x2, y2) or (x, y, w, h)) x1y1x2y2: select the coordinate format return: iou: computed iou """ if not x1y1x2y2: # Transform from center and width to exact coordinates b1_x1, b1_x2 = box1[:, 0] - box1[:, 2] / 2, box1[:, 0] + box1[:, 2] / 2 b1_y1, b1_y2 = box1[:, 1] - box1[:, 3] / 2, box1[:, 1] + box1[:, 3] / 2 b2_x1, b2_x2 = box2[:, 0] - box2[:, 2] / 2, box2[:, 0] + box2[:, 2] / 2 b2_y1, b2_y2 = box2[:, 1] - box2[:, 3] / 2, box2[:, 1] + box2[:, 3] / 2 else: # Get the coordinates of bounding boxes b1_x1, b1_y1, b1_x2, b1_y2 = box1[:, 0], box1[:, 1], box1[:, 2], box1[:, 3] b2_x1, b2_y1, b2_x2, b2_y2 = box2[:, 0], box2[:, 1], box2[:, 2], box2[:, 3] # Get the coordinates of the intersection rectangle inter_rect_x1 = np.maximum(b1_x1, b2_x1) inter_rect_y1 = np.maximum(b1_y1, b2_y1) inter_rect_x2 = np.minimum(b1_x2, b2_x2) inter_rect_y2 = np.minimum(b1_y2, b2_y2) # Intersection area inter_area = np.clip(inter_rect_x2 - inter_rect_x1 + 1, 0, None) * \ np.clip(inter_rect_y2 - inter_rect_y1 + 1, 0, None) # Union Area b1_area = (b1_x2 - b1_x1 + 1) * (b1_y2 - b1_y1 + 1) b2_area = (b2_x2 - b2_x1 + 1) * (b2_y2 - b2_y1 + 1) iou = inter_area / (b1_area + b2_area - inter_area + 1e-16) return iou def non_max_suppression(self, prediction, origin_h, origin_w, conf_thres=0.5, nms_thres=0.4): """ description: Removes detections with lower object confidence score than 'conf_thres' and performs Non-Maximum Suppression to further filter detections. param: prediction: detections, (x1, y1, x2, y2, conf, cls_id) origin_h: original image height origin_w: original image width conf_thres: a confidence threshold to filter detections nms_thres: a iou threshold to filter detections return: boxes: output after nms with the shape (x1, y1, x2, y2, conf, cls_id) """ # Get the boxes that score > CONF_THRESH boxes = prediction[prediction[:, 4] >= conf_thres] # Trandform bbox from [center_x, center_y, w, h] to [x1, y1, x2, y2] boxes[:, :4] = self.xywh2xyxy(origin_h, origin_w, boxes[:, :4]) # clip the coordinates boxes[:, 0] = np.clip(boxes[:, 0], 0, origin_w -1) boxes[:, 2] = np.clip(boxes[:, 2], 0, origin_w -1) boxes[:, 1] = np.clip(boxes[:, 1], 0, origin_h -1) boxes[:, 3] = np.clip(boxes[:, 3], 0, origin_h -1) # Object confidence confs = boxes[:, 4] # Sort by the confs boxes = boxes[np.argsort(-confs)] # Perform non-maximum suppression keep_boxes = [] while boxes.shape[0]: large_overlap = self.bbox_iou(np.expand_dims(boxes[0, :4], 0), boxes[:, :4]) > nms_thres label_match = boxes[0, -1] == boxes[:, -1] # Indices of boxes with lower confidence scores, large IOUs and matching labels invalid = large_overlap & label_match keep_boxes += [boxes[0]] boxes = boxes[~invalid] boxes = np.stack(keep_boxes, 0) if len(keep_boxes) else np.array([]) return boxes # class inferThread(threading.Thread): # def __init__(self, yolov5_wrapper, image_path_batch): # threading.Thread.__init__(self) # self.yolov5_wrapper = yolov5_wrapper # self.image_path_batch = image_path_batch # def run(self): # batch_image_raw, use_time = self.yolov5_wrapper.infer(self.yolov5_wrapper.get_raw_image(self.image_path_batch)) # for i, img_path in enumerate(self.image_path_batch): # parent, filename = os.path.split(img_path) # save_name = os.path.join('output', filename) # # Save image # cv2.imwrite(save_name, batch_image_raw[i]) # print('input->{}, time->{:.2f}ms, saving into output/'.format(self.image_path_batch, use_time * 1000)) class warmUpThread(threading.Thread): def __init__(self, yolov5_wrapper): threading.Thread.__init__(self) self.yolov5_wrapper = yolov5_wrapper def run(self): batch_image_raw, use_time = self.yolov5_wrapper.infer(self.yolov5_wrapper.get_raw_image_zeros()) print('warm_up->{}, time->{:.2f}ms'.format(batch_image_raw[0].shape, use_time * 1000)) # def get_max_area(img , boxs, result_classid, result_scores): # max_area = 0 # cls_id = 10 # score_1=0 # for box, class_id, score in zip(boxs, result_classid, result_scores): # x1, y1, x2, y2 = box # area = (x2-x1)*(y2-y1) # if area > 100: # cv2.rectangle(img, (int(x1), int(y1)), (int(x2), int(y2)), (255, 0, 0), # thickness=3, lineType=cv2.LINE_AA) # if area > max_area: # max_area = int(area) # cls_id = int(class_id) # score_1=score # return cls_id, max_area, score_1 def get_max_area(boxs, result_classid , result_scores): max_area = 0 cls_id = 0 _score = 0 max_box=[] for box, class_id ,score in zip(boxs, result_classid,result_scores): x1, y1, x2, y2 = box area = (y2-y1)*(y2-y1) if area > max_area: max_area = int(area) cls_id = int(class_id) max_box = box _score = score return cls_id, max_area, max_box ,_score def draw_rerectangle(frame,box,cls,score,area): x1, y1, x2, y2 = box cv2.rectangle(frame, (int(x1), int(y1)), (int(x2), int(y2)), (255, 0, 0), thickness=3, lineType=cv2.LINE_AA) cv2.putText(frame, str("cls_id:%s" % cls), (20, 60), cv2.FONT_HERSHEY_SIMPLEX, 0.5, (0, 255, 0), 2) cv2.putText(frame, str("area:%d" % area), (20, 120), cv2.FONT_HERSHEY_SIMPLEX, 0.5, (0, 255, 0), 2) cv2.putText(frame, str("score:%f" % score), (20, 90), cv2.FONT_HERSHEY_SIMPLEX, 0.5, (0, 255, 0), 2) if __name__ == "__main__": # load custom plugin and engine PLUGIN_LIBRARY = "/media/jian/My Passport/资料/car/ai_control-FULL-3.0(14)/ai_control-FULL-3.0/src/opencv_detection/scripts/small_yolo/libmyplugins.so" engine_file_path = "/media/jian/My Passport/资料/car/ai_control-FULL-3.0(14)/ai_control-FULL-3.0/src/opencv_detection/scripts/small_yolo/small.engine" if len(sys.argv) > 1: engine_file_path = sys.argv[1] if len(sys.argv) > 2: PLUGIN_LIBRARY = sys.argv[2] ctypes.CDLL(PLUGIN_LIBRARY) # load coco labels categories = ["danger","side_walk","speed","speed_limit","turn_left","turn_right","lane_change"] # a YoLov5TRT instance yolov5_wrapper = YoLov5TRT(engine_file_path) cap = cv2.VideoCapture(0,cv2.CAP_V4L2) try: while 1: ret, img = cap.read() if not ret: continue image = cv2.resize(img, (320, 240), interpolation=cv2.INTER_AREA) #w*h # print(image.shape) # frame = image[:224, 96:320, :] #h*w result_boxes, result_scores, result_classid, use_time = yolov5_wrapper.infer( image) cls_id = 10 area = 0 if (len(result_classid) != 0): cls_id, area, score = get_max_area(image,result_boxes, result_classid,result_scores) cv2.putText(image, str("cls_id:%s" % categories[cls_id]), (20, 100), cv2.FONT_HERSHEY_SIMPLEX, 0.5, (0, 255, 0), 2) cv2.putText(image, str("area:%d" % area), (20, 160), cv2.FONT_HERSHEY_SIMPLEX, 0.5, (0, 255, 0), 2) cv2.putText(image, str("score:%f" % score), (20, 130), cv2.FONT_HERSHEY_SIMPLEX, 0.5, (0, 255, 0), 2) cv2.imshow("frame", image) if cv2.waitKey(1) & 0xFF == ord('q'): break finally: # destroy the instance cap.release() cv2.destroyAllWindows() yolov5_wrapper.destroy()

filetype

from data import COCODetection, get_label_map, MEANS, COLORS from yolact import Yolact from utils.augmentations import BaseTransform, FastBaseTransform, Resize from utils.functions import MovingAverage, ProgressBar from layers.box_utils import jaccard, center_size, mask_iou from utils import timer from utils.functions import SavePath from layers.output_utils import postprocess, undo_image_transformation import pycocotools from data import cfg, set_cfg, set_dataset import numpy as np import torch import torch.backends.cudnn as cudnn from torch.autograd import Variable import argparse import time import random import cProfile import pickle import json import os from collections import defaultdict from pathlib import Path from collections import OrderedDict from PIL import Image import matplotlib.pyplot as plt import cv2 def str2bool(v): if v.lower() in ('yes', 'true', 't', 'y', '1'): return True elif v.lower() in ('no', 'false', 'f', 'n', '0'): return False else: raise argparse.ArgumentTypeError('Boolean value expected.') def parse_args(argv=None): parser = argparse.ArgumentParser( description='YOLACT COCO Evaluation') parser.add_argument('--trained_model', default='weights/yolact_base_105_101798_interrupt.pth', type=str, help='Trained state_dict file path to open. If "interrupt", this will open the interrupt file.') parser.add_argument('--top_k', default=5, type=int, help='Further restrict the number of predictions to parse') parser.add_argument('--cuda', default=True, type=str2bool, help='Use cuda to evaulate model') parser.add_argument('--fast_nms', default=True, type=str2bool, help='Whether to use a faster, but not entirely correct version of NMS.') parser.add_argument('--cross_class_nms', default=False, type=str2bool, help='Whether compute NMS cross-class or per-class.') parser.add_argument('--display_masks', default=True, type=str2bool, help='Whether or not to display masks over bounding boxes') parser.add_argument('--display_bboxes', default=True, type=str2bool, help='Whether or not to display bboxes around masks') parser.add_argument('--display_text', default=True, type=str2bool, help='Whether or not to display text (class [score])') parser.add_argument('--display_scores', default=True, type=str2bool, help='Whether or not to display scores in addition to classes') parser.add_argument('--display', dest='display', action='store_true', help='Display qualitative results instead of quantitative ones.') parser.add_argument('--shuffle', dest='shuffle', action='store_true', help='Shuffles the images when displaying them. Doesn\'t have much of an effect when display is off though.') parser.add_argument('--ap_data_file', default='results/ap_data.pkl', type=str, help='In quantitative mode, the file to save detections before calculating mAP.') parser.add_argument('--resume', dest='resume', action='store_true', help='If display not set, this resumes mAP calculations from the ap_data_file.') parser.add_argument('--max_images', default=-1, type=int, help='The maximum number of images from the dataset to consider. Use -1 for all.') parser.add_argument('--output_coco_json', dest='output_coco_json', action='store_true', help='If display is not set, instead of processing IoU values, this just dumps detections into the coco json file.') parser.add_argument('--bbox_det_file', default='results/bbox_detections.json', type=str, help='The output file for coco bbox results if --coco_results is set.') parser.add_argument('--mask_det_file', default='results/mask_detections.json', type=str, help='The output file for coco mask results if --coco_results is set.') parser.add_argument('--config', default=None, help='The config object to use.') parser.add_argument('--output_web_json', dest='output_web_json', action='store_true', help='If display is not set, instead of processing IoU values, this dumps detections for usage with the detections viewer web thingy.') parser.add_argument('--web_det_path', default='web/dets/', type=str, help='If output_web_json is set, this is the path to dump detections into.') parser.add_argument('--no_bar', dest='no_bar', action='store_true', help='Do not output the status bar. This is useful for when piping to a file.') parser.add_argument('--display_lincomb', default=False, type=str2bool, help='If the config uses lincomb masks, output a visualization of how those masks are created.') parser.add_argument('--benchmark', default=False, dest='benchmark', action='store_true', help='Equivalent to running display mode but without displaying an image.') parser.add_argument('--no_sort', default=False, dest='no_sort', action='store_true', help='Do not sort images by hashed image ID.') parser.add_argument('--seed', default=None, type=int, help='The seed to pass into random.seed. Note: this is only really for the shuffle and does not (I think) affect cuda stuff.') parser.add_argument('--mask_proto_debug', default=False, dest='mask_proto_debug', action='store_true', help='Outputs stuff for scripts/compute_mask.py.') parser.add_argument('--no_crop', default=False, dest='crop', action='store_false', help='Do not crop output masks with the predicted bounding box.') parser.add_argument('--image', default=None, type=str, help='A path to an image to use for display.') parser.add_argument('--images', default='E:/yolact-master/coco/images/train2017', type=str, help='Input and output paths separated by a colon.') parser.add_argument('--video', default=None, type=str, help='A path to a video to evaluate on. Passing in a number will use that index webcam.') parser.add_argument('--video_multiframe', default=1, type=int, help='The number of frames to evaluate in parallel to make videos play at higher fps.') parser.add_argument('--score_threshold', default=0.15, type=float, help='Detections with a score under this threshold will not be considered. This currently only works in display mode.') parser.add_argument('--dataset', default=None, type=str, help='If specified, override the dataset specified in the config with this one (example: coco2017_dataset).') parser.add_argument('--detect', default=False, dest='detect', action='store_true', help='Don\'t evauluate the mask branch at all and only do object detection. This only works for --display and --benchmark.') parser.add_argument('--display_fps', default=False, dest='display_fps', action='store_true', help='When displaying / saving video, draw the FPS on the frame') parser.add_argument('--emulate_playback', default=False, dest='emulate_playback', action='store_true', help='When saving a video, emulate the framerate that you\'d get running in real-time mode.') parser.set_defaults(no_bar=False, display=False, resume=False, output_coco_json=False, output_web_json=False, shuffle=False, benchmark=False, no_sort=False, no_hash=False, mask_proto_debug=False, crop=True, detect=False, display_fps=False, emulate_playback=False) global args args = parser.parse_args(argv) if args.output_web_json: args.output_coco_json = True if args.seed is not None: random.seed(args.seed) iou_thresholds = [x / 100 for x in range(50, 100, 5)] coco_cats = {} # Call prep_coco_cats to fill this coco_cats_inv = {} color_cache = defaultdict(lambda: {}) def prep_display(dets_out, img, h, w, undo_transform=True, class_color=False, mask_alpha=0.45, fps_str=''): """ Note: If undo_transform=False then im_h and im_w are allowed to be None. """ if undo_transform: img_numpy = undo_image_transformation(img, w, h) img_gpu = torch.Tensor(img_numpy).cuda() else: img_gpu = img / 255.0 h, w, _ = img.shape with timer.env('Postprocess'): save = cfg.rescore_bbox cfg.rescore_bbox = True t = postprocess(dets_out, w, h, visualize_lincomb = args.display_lincomb, crop_masks = args.crop, score_threshold = args.score_threshold) cfg.rescore_bbox = save with timer.env('Copy'): idx = t[1].argsort(0, descending=True)[:args.top_k] if cfg.eval_mask_branch: # Masks are drawn on the GPU, so don't copy masks = t[3][idx] classes, scores, boxes = [x[idx].cpu().numpy() for x in t[:3]] num_dets_to_consider = min(args.top_k, classes.shape[0]) for j in range(num_dets_to_consider): if scores[j] < args.score_threshold: num_dets_to_consider = j break # Quick and dirty lambda for selecting the color for a particular index # Also keeps track of a per-gpu color cache for maximum speed def get_color(j, on_gpu=None): global color_cache color_idx = (classes[j] * 5 if class_color else j * 5) % len(COLORS) if on_gpu is not None and color_idx in color_cache[on_gpu]: return color_cache[on_gpu][color_idx] else: color = COLORS[color_idx] if not undo_transform: # The image might come in as RGB or BRG, depending color = (color[2], color[1], color[0]) if on_gpu is not None: color = torch.Tensor(color).to(on_gpu).float() / 255. color_cache[on_gpu][color_idx] = color return color # First, draw the masks on the GPU where we can do it really fast # Beware: very fast but possibly unintelligible mask-drawing code ahead # I wish I had access to OpenGL or Vulkan but alas, I guess Pytorch tensor operations will have to suffice if args.display_masks and cfg.eval_mask_branch and num_dets_to_consider > 0: # After this, mask is of size [num_dets, h, w, 1] masks = masks[:num_dets_to_consider, :, :, None] # Prepare the RGB images for each mask given their color (size [num_dets, h, w, 1]) colors = torch.cat([get_color(j, on_gpu=img_gpu.device.index).view(1, 1, 1, 3) for j in range(num_dets_to_consider)], dim=0) masks_color = masks.repeat(1, 1, 1, 3) * colors * mask_alpha # This is 1 everywhere except for 1-mask_alpha where the mask is inv_alph_masks = masks * (-mask_alpha) + 1 # I did the math for this on pen and paper. This whole block should be equivalent to: # for j in range(num_dets_to_consider): # img_gpu = img_gpu * inv_alph_masks[j] + masks_color[j] masks_color_summand = masks_color[0] if num_dets_to_consider > 1: inv_alph_cumul = inv_alph_masks[:(num_dets_to_consider-1)].cumprod(dim=0) masks_color_cumul = masks_color[1:] * inv_alph_cumul masks_color_summand += masks_color_cumul.sum(dim=0) img_gpu = img_gpu * inv_alph_masks.prod(dim=0) + masks_color_summand if args.display_fps: # Draw the box for the fps on the GPU font_face = cv2.FONT_HERSHEY_DUPLEX font_scale = 0.6 font_thickness = 1 text_w, text_h = cv2.getTextSize(fps_str, font_face, font_scale, font_thickness)[0] img_gpu[0:text_h+8, 0:text_w+8] *= 0.6 # 1 - Box alpha # Then draw the stuff that needs to be done on the cpu # Note, make sure this is a uint8 tensor or opencv will not anti alias text for whatever reason img_numpy = (img_gpu * 255).byte().cpu().numpy() if args.display_fps: # Draw the text on the CPU text_pt = (4, text_h + 2) text_color = [255, 255, 255] cv2.putText(img_numpy, fps_str, text_pt, font_face, font_scale, text_color, font_thickness, cv2.LINE_AA) if num_dets_to_consider == 0: return img_numpy if args.display_text or args.display_bboxes: for j in reversed(range(num_dets_to_consider)): x1, y1, x2, y2 = boxes[j, :] color = get_color(j) score = scores[j] if args.display_bboxes: cv2.rectangle(img_numpy, (x1, y1), (x2, y2), color, 1) if args.display_text: _class = cfg.dataset.class_names[classes[j]] text_str = '%s: %.2f' % (_class, score) if args.display_scores else _class font_face = cv2.FONT_HERSHEY_DUPLEX font_scale = 0.6 font_thickness = 1 text_w, text_h = cv2.getTextSize(text_str, font_face, font_scale, font_thickness)[0] text_pt = (x1, y1 - 3) text_color = [255, 255, 255] cv2.rectangle(img_numpy, (x1, y1), (x1 + text_w, y1 - text_h - 4), color, -1) cv2.putText(img_numpy, text_str, text_pt, font_face, font_scale, text_color, font_thickness, cv2.LINE_AA) return img_numpy def prep_benchmark(dets_out, h, w): with timer.env('Postprocess'): t = postprocess(dets_out, w, h, crop_masks=args.crop, score_threshold=args.score_threshold) with timer.env('Copy'): classes, scores, boxes, masks = [x[:args.top_k] for x in t] if isinstance(scores, list): box_scores = scores[0].cpu().numpy() mask_scores = scores[1].cpu().numpy() else: scores = scores.cpu().numpy() classes = classes.cpu().numpy() boxes = boxes.cpu().numpy() masks = masks.cpu().numpy() with timer.env('Sync'): # Just in case torch.cuda.synchronize() def prep_coco_cats(): """ Prepare inverted table for category id lookup given a coco cats object. """ for coco_cat_id, transformed_cat_id_p1 in get_label_map().items(): transformed_cat_id = transformed_cat_id_p1 - 1 coco_cats[transformed_cat_id] = coco_cat_id coco_cats_inv[coco_cat_id] = transformed_cat_id def get_coco_cat(transformed_cat_id): """ transformed_cat_id is [0,80) as indices in cfg.dataset.class_names """ return coco_cats[transformed_cat_id] def get_transformed_cat(coco_cat_id): """ transformed_cat_id is [0,80) as indices in cfg.dataset.class_names """ return coco_cats_inv[coco_cat_id] class Detections: def __init__(self): self.bbox_data = [] self.mask_data = [] def add_bbox(self, image_id:int, category_id:int, bbox:list, score:float): """ Note that bbox should be a list or tuple of (x1, y1, x2, y2) """ bbox = [bbox[0], bbox[1], bbox[2]-bbox[0], bbox[3]-bbox[1]] # Round to the nearest 10th to avoid huge file sizes, as COCO suggests bbox = [round(float(x)*10)/10 for x in bbox] self.bbox_data.append({ 'image_id': int(image_id), 'category_id': get_coco_cat(int(category_id)), 'bbox': bbox, 'score': float(score) }) def add_mask(self, image_id:int, category_id:int, segmentation:np.ndarray, score:float): """ The segmentation should be the full mask, the size of the image and with size [h, w]. """ rle = pycocotools.mask.encode(np.asfortranarray(segmentation.astype(np.uint8))) rle['counts'] = rle['counts'].decode('ascii') # json.dump doesn't like bytes strings self.mask_data.append({ 'image_id': int(image_id), 'category_id': get_coco_cat(int(category_id)), 'segmentation': rle, 'score': float(score) }) def dump(self): dump_arguments = [ (self.bbox_data, args.bbox_det_file), (self.mask_data, args.mask_det_file) ] for data, path in dump_arguments: with open(path, 'w') as f: json.dump(data, f) def dump_web(self): """ Dumps it in the format for my web app. Warning: bad code ahead! """ config_outs = ['preserve_aspect_ratio', 'use_prediction_module', 'use_yolo_regressors', 'use_prediction_matching', 'train_masks'] output = { 'info' : { 'Config': {key: getattr(cfg, key) for key in config_outs}, } } image_ids = list(set([x['image_id'] for x in self.bbox_data])) image_ids.sort() image_lookup = {_id: idx for idx, _id in enumerate(image_ids)} output['images'] = [{'image_id': image_id, 'dets': []} for image_id in image_ids] # These should already be sorted by score with the way prep_metrics works. for bbox, mask in zip(self.bbox_data, self.mask_data): image_obj = output['images'][image_lookup[bbox['image_id']]] image_obj['dets'].append({ 'score': bbox['score'], 'bbox': bbox['bbox'], 'category': cfg.dataset.class_names[get_transformed_cat(bbox['category_id'])], 'mask': mask['segmentation'], }) with open(os.path.join(args.web_det_path, '%s.json' % cfg.name), 'w') as f: json.dump(output, f) def _mask_iou(mask1, mask2, iscrowd=False): with timer.env('Mask IoU'): ret = mask_iou(mask1, mask2, iscrowd) return ret.cpu() def _bbox_iou(bbox1, bbox2, iscrowd=False): with timer.env('BBox IoU'): ret = jaccard(bbox1, bbox2, iscrowd) return ret.cpu() def prep_metrics(ap_data, dets, img, gt, gt_masks, h, w, num_crowd, image_id, detections:Detections=None): """ Returns a list of APs for this image, with each element being for a class """ if not args.output_coco_json: with timer.env('Prepare gt'): gt_boxes = torch.Tensor(gt[:, :4]) gt_boxes[:, [0, 2]] *= w gt_boxes[:, [1, 3]] *= h gt_classes = list(gt[:, 4].astype(int)) gt_masks = torch.Tensor(gt_masks).view(-1, h*w) if num_crowd > 0: split = lambda x: (x[-num_crowd:], x[:-num_crowd]) crowd_boxes , gt_boxes = split(gt_boxes) crowd_masks , gt_masks = split(gt_masks) crowd_classes, gt_classes = split(gt_classes) with timer.env('Postprocess'): classes, scores, boxes, masks = postprocess(dets, w, h, crop_masks=args.crop, score_threshold=args.score_threshold) if classes.size(0) == 0: return classes = list(classes.cpu().numpy().astype(int)) if isinstance(scores, list): box_scores = list(scores[0].cpu().numpy().astype(float)) mask_scores = list(scores[1].cpu().numpy().astype(float)) else: scores = list(scores.cpu().numpy().astype(float)) box_scores = scores mask_scores = scores masks = masks.view(-1, h*w).cuda() boxes = boxes.cuda() if args.output_coco_json: with timer.env('JSON Output'): boxes = boxes.cpu().numpy() masks = masks.view(-1, h, w).cpu().numpy() for i in range(masks.shape[0]): # Make sure that the bounding box actually makes sense and a mask was produced if (boxes[i, 3] - boxes[i, 1]) * (boxes[i, 2] - boxes[i, 0]) > 0: detections.add_bbox(image_id, classes[i], boxes[i,:], box_scores[i]) detections.add_mask(image_id, classes[i], masks[i,:,:], mask_scores[i]) return with timer.env('Eval Setup'): num_pred = len(classes) num_gt = len(gt_classes) mask_iou_cache = _mask_iou(masks, gt_masks) bbox_iou_cache = _bbox_iou(boxes.float(), gt_boxes.float()) if num_crowd > 0: crowd_mask_iou_cache = _mask_iou(masks, crowd_masks, iscrowd=True) crowd_bbox_iou_cache = _bbox_iou(boxes.float(), crowd_boxes.float(), iscrowd=True) else: crowd_mask_iou_cache = None crowd_bbox_iou_cache = None box_indices = sorted(range(num_pred), key=lambda i: -box_scores[i]) mask_indices = sorted(box_indices, key=lambda i: -mask_scores[i]) iou_types = [ ('box', lambda i,j: bbox_iou_cache[i, j].item(), lambda i,j: crowd_bbox_iou_cache[i,j].item(), lambda i: box_scores[i], box_indices), ('mask', lambda i,j: mask_iou_cache[i, j].item(), lambda i,j: crowd_mask_iou_cache[i,j].item(), lambda i: mask_scores[i], mask_indices) ] timer.start('Main loop') for _class in set(classes + gt_classes): ap_per_iou = [] num_gt_for_class = sum([1 for x in gt_classes if x == _class]) for iouIdx in range(len(iou_thresholds)): iou_threshold = iou_thresholds[iouIdx] for iou_type, iou_func, crowd_func, score_func, indices in iou_types: gt_used = [False] * len(gt_classes) ap_obj = ap_data[iou_type][iouIdx][_class] ap_obj.add_gt_positives(num_gt_for_class) for i in indices: if classes[i] != _class: continue max_iou_found = iou_threshold max_match_idx = -1 for j in range(num_gt): if gt_used[j] or gt_classes[j] != _class: continue iou = iou_func(i, j) if iou > max_iou_found: max_iou_found = iou max_match_idx = j if max_match_idx >= 0: gt_used[max_match_idx] = True ap_obj.push(score_func(i), True) else: # If the detection matches a crowd, we can just ignore it matched_crowd = False if num_crowd > 0: for j in range(len(crowd_classes)): if crowd_classes[j] != _class: continue iou = crowd_func(i, j) if iou > iou_threshold: matched_crowd = True break # All this crowd code so that we can make sure that our eval code gives the # same result as COCOEval. There aren't even that many crowd annotations to # begin with, but accuracy is of the utmost importance. if not matched_crowd: ap_obj.push(score_func(i), False) timer.stop('Main loop') class APDataObject: """ Stores all the information necessary to calculate the AP for one IoU and one class. Note: I type annotated this because why not. """ def __init__(self): self.data_points = [] self.num_gt_positives = 0 def push(self, score:float, is_true:bool): self.data_points.append((score, is_true)) def add_gt_positives(self, num_positives:int): """ Call this once per image. """ self.num_gt_positives += num_positives def is_empty(self) -> bool: return len(self.data_points) == 0 and self.num_gt_positives == 0 def get_ap(self) -> float: """ Warning: result not cached. """ if self.num_gt_positives == 0: return 0 # Sort descending by score self.data_points.sort(key=lambda x: -x[0]) precisions = [] recalls = [] num_true = 0 num_false = 0 # Compute the precision-recall curve. The x axis is recalls and the y axis precisions. for datum in self.data_points: # datum[1] is whether the detection a true or false positive if datum[1]: num_true += 1 else: num_false += 1 precision = num_true / (num_true + num_false) recall = num_true / self.num_gt_positives precisions.append(precision) recalls.append(recall) # Smooth the curve by computing [max(precisions[i:]) for i in range(len(precisions))] # Basically, remove any temporary dips from the curve. # At least that's what I think, idk. COCOEval did it so I do too. for i in range(len(precisions)-1, 0, -1): if precisions[i] > precisions[i-1]: precisions[i-1] = precisions[i] # Compute the integral of precision(recall) d_recall from recall=0->1 using fixed-length riemann summation with 101 bars. y_range = [0] * 101 # idx 0 is recall == 0.0 and idx 100 is recall == 1.00 x_range = np.array([x / 100 for x in range(101)]) recalls = np.array(recalls) # I realize this is weird, but all it does is find the nearest precision(x) for a given x in x_range. # Basically, if the closest recall we have to 0.01 is 0.009 this sets precision(0.01) = precision(0.009). # I approximate the integral this way, because that's how COCOEval does it. indices = np.searchsorted(recalls, x_range, side='left') for bar_idx, precision_idx in enumerate(indices): if precision_idx < len(precisions): y_range[bar_idx] = precisions[precision_idx] # Finally compute the riemann sum to get our integral. # avg([precision(x) for x in 0:0.01:1]) return sum(y_range) / len(y_range) def badhash(x): """ Just a quick and dirty hash function for doing a deterministic shuffle based on image_id. Source: https://stackoverflow.com/questions/664014/what-integer-hash-function-are-good-that-accepts-an-integer-hash-key """ x = (((x >> 16) ^ x) * 0x045d9f3b) & 0xFFFFFFFF x = (((x >> 16) ^ x) * 0x045d9f3b) & 0xFFFFFFFF x = ((x >> 16) ^ x) & 0xFFFFFFFF return x def evalimage(net:Yolact, path:str, save_path:str=None): frame = torch.from_numpy(cv2.imread(path)).cuda().float() batch = FastBaseTransform()(frame.unsqueeze(0)) preds = net(batch) img_numpy = prep_display(preds, frame, None, None, undo_transform=False) if save_path is None: img_numpy = img_numpy[:, :, (2, 1, 0)] if save_path is None: plt.imshow(img_numpy) plt.title(path) plt.show() else: cv2.imwrite(save_path, img_numpy) def evalimages(net:Yolact, input_folder:str, output_folder:str): if not os.path.exists(output_folder): os.mkdir(output_folder) print() for p in Path(input_folder).glob('*'): path = str(p) name = os.path.basename(path) name = '.'.join(name.split('.')[:-1]) + '.png' out_path = os.path.join(output_folder, name) evalimage(net, path, out_path) print(path + ' -> ' + out_path) print('Done.') from multiprocessing.pool import ThreadPool from queue import Queue class CustomDataParallel(torch.nn.DataParallel): """ A Custom Data Parallel class that properly gathers lists of dictionaries. """ def gather(self, outputs, output_device): # Note that I don't actually want to convert everything to the output_device return sum(outputs, []) def evalvideo(net:Yolact, path:str, out_path:str=None): # If the path is a digit, parse it as a webcam index is_webcam = path.isdigit() # If the input image size is constant, this make things faster (hence why we can use it in a video setting). cudnn.benchmark = True if is_webcam: vid = cv2.VideoCapture(int(path)) else: vid = cv2.VideoCapture(path) if not vid.isOpened(): print('Could not open video "%s"' % path) exit(-1) target_fps = round(vid.get(cv2.CAP_PROP_FPS)) frame_width = round(vid.get(cv2.CAP_PROP_FRAME_WIDTH)) frame_height = round(vid.get(cv2.CAP_PROP_FRAME_HEIGHT)) if is_webcam: num_frames = float('inf') else: num_frames = round(vid.get(cv2.CAP_PROP_FRAME_COUNT)) net = CustomDataParallel(net).cuda() transform = torch.nn.DataParallel(FastBaseTransform()).cuda() frame_times = MovingAverage(100) fps = 0 frame_time_target = 1 / target_fps running = True fps_str = '' vid_done = False frames_displayed = 0 if out_path is not None: out = cv2.VideoWriter(out_path, cv2.VideoWriter_fourcc(*"mp4v"), target_fps, (frame_width, frame_height)) def cleanup_and_exit(): print() pool.terminate() vid.release() if out_path is not None: out.release() cv2.destroyAllWindows() exit() def get_next_frame(vid): frames = [] for idx in range(args.video_multiframe): frame = vid.read()[1] if frame is None: return frames frames.append(frame) return frames def transform_frame(frames): with torch.no_grad(): frames = [torch.from_numpy(frame).cuda().float() for frame in frames] return frames, transform(torch.stack(frames, 0)) def eval_network(inp): with torch.no_grad(): frames, imgs = inp num_extra = 0 while imgs.size(0) < args.video_multiframe: imgs = torch.cat([imgs, imgs[0].unsqueeze(0)], dim=0) num_extra += 1 out = net(imgs) if num_extra > 0: out = out[:-num_extra] return frames, out def prep_frame(inp, fps_str): with torch.no_grad(): frame, preds = inp return prep_display(preds, frame, None, None, undo_transform=False, class_color=True, fps_str=fps_str) frame_buffer = Queue() video_fps = 0 # All this timing code to make sure that def play_video(): try: nonlocal frame_buffer, running, video_fps, is_webcam, num_frames, frames_displayed, vid_done video_frame_times = MovingAverage(100) frame_time_stabilizer = frame_time_target last_time = None stabilizer_step = 0.0005 progress_bar = ProgressBar(30, num_frames) while running: frame_time_start = time.time() if not frame_buffer.empty(): next_time = time.time() if last_time is not None: video_frame_times.add(next_time - last_time) video_fps = 1 / video_frame_times.get_avg() if out_path is None: cv2.imshow(path, frame_buffer.get()) else: out.write(frame_buffer.get()) frames_displayed += 1 last_time = next_time if out_path is not None: if video_frame_times.get_avg() == 0: fps = 0 else: fps = 1 / video_frame_times.get_avg() progress = frames_displayed / num_frames * 100 progress_bar.set_val(frames_displayed) print('\rProcessing Frames %s %6d / %6d (%5.2f%%) %5.2f fps ' % (repr(progress_bar), frames_displayed, num_frames, progress, fps), end='') # This is split because you don't want savevideo to require cv2 display functionality (see #197) if out_path is None and cv2.waitKey(1) == 27: # Press Escape to close running = False if not (frames_displayed < num_frames): running = False if not vid_done: buffer_size = frame_buffer.qsize() if buffer_size < args.video_multiframe: frame_time_stabilizer += stabilizer_step elif buffer_size > args.video_multiframe: frame_time_stabilizer -= stabilizer_step if frame_time_stabilizer < 0: frame_time_stabilizer = 0 new_target = frame_time_stabilizer if is_webcam else max(frame_time_stabilizer, frame_time_target) else: new_target = frame_time_target next_frame_target = max(2 * new_target - video_frame_times.get_avg(), 0) target_time = frame_time_start + next_frame_target - 0.001 # Let's just subtract a millisecond to be safe if out_path is None or args.emulate_playback: # This gives more accurate timing than if sleeping the whole amount at once while time.time() < target_time: time.sleep(0.001) else: # Let's not starve the main thread, now time.sleep(0.001) except: # See issue #197 for why this is necessary import traceback traceback.print_exc() extract_frame = lambda x, i: (x[0][i] if x[1][i]['detection'] is None else x[0][i].to(x[1][i]['detection']['box'].device), [x[1][i]]) # Prime the network on the first frame because I do some thread unsafe things otherwise print('Initializing model... ', end='') first_batch = eval_network(transform_frame(get_next_frame(vid))) print('Done.') # For each frame the sequence of functions it needs to go through to be processed (in reversed order) sequence = [prep_frame, eval_network, transform_frame] pool = ThreadPool(processes=len(sequence) + args.video_multiframe + 2) pool.apply_async(play_video) active_frames = [{'value': extract_frame(first_batch, i), 'idx': 0} for i in range(len(first_batch[0]))] print() if out_path is None: print('Press Escape to close.') try: while vid.isOpened() and running: # Hard limit on frames in buffer so we don't run out of memory >.> while frame_buffer.qsize() > 100: time.sleep(0.001) start_time = time.time() # Start loading the next frames from the disk if not vid_done: next_frames = pool.apply_async(get_next_frame, args=(vid,)) else: next_frames = None if not (vid_done and len(active_frames) == 0): # For each frame in our active processing queue, dispatch a job # for that frame using the current function in the sequence for frame in active_frames: _args = [frame['value']] if frame['idx'] == 0: _args.append(fps_str) frame['value'] = pool.apply_async(sequence[frame['idx']], args=_args) # For each frame whose job was the last in the sequence (i.e. for all final outputs) for frame in active_frames: if frame['idx'] == 0: frame_buffer.put(frame['value'].get()) # Remove the finished frames from the processing queue active_frames = [x for x in active_frames if x['idx'] > 0] # Finish evaluating every frame in the processing queue and advanced their position in the sequence for frame in list(reversed(active_frames)): frame['value'] = frame['value'].get() frame['idx'] -= 1 if frame['idx'] == 0: # Split this up into individual threads for prep_frame since it doesn't support batch size active_frames += [{'value': extract_frame(frame['value'], i), 'idx': 0} for i in range(1, len(frame['value'][0]))] frame['value'] = extract_frame(frame['value'], 0) # Finish loading in the next frames and add them to the processing queue if next_frames is not None: frames = next_frames.get() if len(frames) == 0: vid_done = True else: active_frames.append({'value': frames, 'idx': len(sequence)-1}) # Compute FPS frame_times.add(time.time() - start_time) fps = args.video_multiframe / frame_times.get_avg() else: fps = 0 fps_str = 'Processing FPS: %.2f | Video Playback FPS: %.2f | Frames in Buffer: %d' % (fps, video_fps, frame_buffer.qsize()) if not args.display_fps: print('\r' + fps_str + ' ', end='') except KeyboardInterrupt: print('\nStopping...') cleanup_and_exit() def evaluate(net:Yolact, dataset, train_mode=False): net.detect.use_fast_nms = args.fast_nms net.detect.use_cross_class_nms = args.cross_class_nms cfg.mask_proto_debug = args.mask_proto_debug # TODO Currently we do not support Fast Mask Re-scroing in evalimage, evalimages, and evalvideo if args.image is not None: if ':' in args.image: inp, out = args.image.split(':') evalimage(net, inp, out) else: evalimage(net, args.image) return elif args.images is not None: inp, out = args.images.split('E:/yolact-master/coco/images/train2017: E:/yolact-master/results/output') evalimages(net, inp, out) return elif args.video is not None: if ':' in args.video: inp, out = args.video.split(':') evalvideo(net, inp, out) else: evalvideo(net, args.video) return frame_times = MovingAverage() dataset_size = len(dataset) if args.max_images < 0 else min(args.max_images, len(dataset)) progress_bar = ProgressBar(30, dataset_size) print() if not args.display and not args.benchmark: # For each class and iou, stores tuples (score, isPositive) # Index ap_data[type][iouIdx][classIdx] ap_data = { 'box' : [[APDataObject() for _ in cfg.dataset.class_names] for _ in iou_thresholds], 'mask': [[APDataObject() for _ in cfg.dataset.class_names] for _ in iou_thresholds] } detections = Detections() else: timer.disable('Load Data') dataset_indices = list(range(len(dataset))) if args.shuffle: random.shuffle(dataset_indices) elif not args.no_sort: # Do a deterministic shuffle based on the image ids # # I do this because on python 3.5 dictionary key order is *random*, while in 3.6 it's # the order of insertion. That means on python 3.6, the images come in the order they are in # in the annotations file. For some reason, the first images in the annotations file are # the hardest. To combat this, I use a hard-coded hash function based on the image ids # to shuffle the indices we use. That way, no matter what python version or how pycocotools # handles the data, we get the same result every time. hashed = [badhash(x) for x in dataset.ids] dataset_indices.sort(key=lambda x: hashed[x]) dataset_indices = dataset_indices[:dataset_size] try: # Main eval loop for it, image_idx in enumerate(dataset_indices): timer.reset() with timer.env('Load Data'): img, gt, gt_masks, h, w, num_crowd = dataset.pull_item(image_idx) # Test flag, do not upvote if cfg.mask_proto_debug: with open('scripts/info.txt', 'w') as f: f.write(str(dataset.ids[image_idx])) np.save('scripts/gt.npy', gt_masks) batch = Variable(img.unsqueeze(0)) if args.cuda: batch = batch.cuda() with timer.env('Network Extra'): preds = net(batch) # Perform the meat of the operation here depending on our mode. if args.display: img_numpy = prep_display(preds, img, h, w) elif args.benchmark: prep_benchmark(preds, h, w) else: prep_metrics(ap_data, preds, img, gt, gt_masks, h, w, num_crowd, dataset.ids[image_idx], detections) # First couple of images take longer because we're constructing the graph. # Since that's technically initialization, don't include those in the FPS calculations. if it > 1: frame_times.add(timer.total_time()) if args.display: if it > 1: print('Avg FPS: %.4f' % (1 / frame_times.get_avg())) plt.imshow(img_numpy) plt.title(str(dataset.ids[image_idx])) plt.show() elif not args.no_bar: if it > 1: fps = 1 / frame_times.get_avg() else: fps = 0 progress = (it+1) / dataset_size * 100 progress_bar.set_val(it+1) print('\rProcessing Images %s %6d / %6d (%5.2f%%) %5.2f fps ' % (repr(progress_bar), it+1, dataset_size, progress, fps), end='') if not args.display and not args.benchmark: print() if args.output_coco_json: print('Dumping detections...') if args.output_web_json: detections.dump_web() else: detections.dump() else: if not train_mode: print('Saving data...') with open(args.ap_data_file, 'wb') as f: pickle.dump(ap_data, f) return calc_map(ap_data) elif args.benchmark: print() print() print('Stats for the last frame:') timer.print_stats() avg_seconds = frame_times.get_avg() print('Average: %5.2f fps, %5.2f ms' % (1 / frame_times.get_avg(), 1000*avg_seconds)) except KeyboardInterrupt: print('Stopping...') def calc_map(ap_data): print('Calculating mAP...') aps = [{'box': [], 'mask': []} for _ in iou_thresholds] for _class in range(len(cfg.dataset.class_names)): for iou_idx in range(len(iou_thresholds)): for iou_type in ('box', 'mask'): ap_obj = ap_data[iou_type][iou_idx][_class] if not ap_obj.is_empty(): aps[iou_idx][iou_type].append(ap_obj.get_ap()) all_maps = {'box': OrderedDict(), 'mask': OrderedDict()} # Looking back at it, this code is really hard to read :/ for iou_type in ('box', 'mask'): all_maps[iou_type]['all'] = 0 # Make this first in the ordereddict for i, threshold in enumerate(iou_thresholds): mAP = sum(aps[i][iou_type]) / len(aps[i][iou_type]) * 100 if len(aps[i][iou_type]) > 0 else 0 all_maps[iou_type][int(threshold*100)] = mAP all_maps[iou_type]['all'] = (sum(all_maps[iou_type].values()) / (len(all_maps[iou_type].values())-1)) print_maps(all_maps) # Put in a prettier format so we can serialize it to json during training all_maps = {k: {j: round(u, 2) for j, u in v.items()} for k, v in all_maps.items()} return all_maps def print_maps(all_maps): # Warning: hacky make_row = lambda vals: (' %5s |' * len(vals)) % tuple(vals) make_sep = lambda n: ('-------+' * n) print() print(make_row([''] + [('.%d ' % x if isinstance(x, int) else x + ' ') for x in all_maps['box'].keys()])) print(make_sep(len(all_maps['box']) + 1)) for iou_type in ('box', 'mask'): print(make_row([iou_type] + ['%.2f' % x if x < 100 else '%.1f' % x for x in all_maps[iou_type].values()])) print(make_sep(len(all_maps['box']) + 1)) print() if __name__ == '__main__': parse_args() if args.config is not None: set_cfg(args.config) if args.trained_model == 'interrupt': args.trained_model = SavePath.get_interrupt('weights/') elif args.trained_model == 'latest': args.trained_model = SavePath.get_latest('weights/', cfg.name) if args.config is None: model_path = SavePath.from_str(args.trained_model) # TODO: Bad practice? Probably want to do a name lookup instead. args.config = model_path.model_name + '_config' print('Config not specified. Parsed %s from the file name.\n' % args.config) set_cfg(args.config) if args.detect: cfg.eval_mask_branch = False if args.dataset is not None: set_dataset(args.dataset) with torch.no_grad(): if not os.path.exists('results'): os.makedirs('results') if args.cuda: cudnn.fastest = True torch.set_default_tensor_type('torch.cuda.FloatTensor') else: torch.set_default_tensor_type('torch.FloatTensor') if args.resume and not args.display: with open(args.ap_data_file, 'rb') as f: ap_data = pickle.load(f) calc_map(ap_data) exit() if args.image is None and args.video is None and args.images is None: dataset = COCODetection(cfg.dataset.valid_images, cfg.dataset.valid_info, transform=BaseTransform(), has_gt=cfg.dataset.has_gt) prep_coco_cats() else: dataset = None print('Loading model...', end='') net = Yolact() net.load_weights(args.trained_model) net.eval() print(' Done.') if args.cuda: net = net.cuda() evaluate(net, dataset) Traceback (most recent call last): File "eval.py", line 1105, in <module> evaluate(net, dataset) File "eval.py", line 884, in evaluate inp, out = args.images.split('E:/yolact-master/coco/images/train2017: E:/yolact-master/results/output') ValueError: not enough values to unpack (expected 2, got 1)

filetype

from random import random import cv2 import numpy as np import onnxruntime as ort import time def plot_one_box(x, img, color=None, label=None, line_thickness=None): """ description: Plots one bounding box on image img, this function comes from YoLov5 project. param: x: a box likes [x1,y1,x2,y2] img: a opencv image object color: color to draw rectangle, such as (0,255,0) label: str line_thickness: int return: no return """ tl = ( line_thickness or round(0.002 * (img.shape[0] + img.shape[1]) / 2) + 1 ) # line/font thickness color = color or [random.randint(0, 255) for _ in range(3)] x = x.squeeze() c1, c2 = (int(x[0]), int(x[1])), (int(x[2]), int(x[3])) cv2.rectangle(img, c1, c2, color, thickness=tl, lineType=cv2.LINE_AA) if label: tf = max(tl - 1, 1) # font thickness t_size = cv2.getTextSize(label, 0, fontScale=tl / 3, thickness=tf)[0] c2 = c1[0] + t_size[0], c1[1] - t_size[1] - 3 cv2.rectangle(img, c1, c2, color, -1, cv2.LINE_AA) # filled cv2.putText( img, label, (c1[0], c1[1] - 2), 0, tl / 3, [225, 255, 255], thickness=tf, lineType=cv2.LINE_AA, ) def _make_grid(nx, ny): xv, yv = np.meshgrid(np.arange(ny), np.arange(nx)) return np.stack((xv, yv), 2).reshape((-1, 2)).astype(np.float32) def cal_outputs(outs, nl, na, model_w, model_h, anchor_grid, stride): row_ind = 0 grid = [np.zeros(1)] * nl for i in range(nl): h, w = int(model_w / stride[i]), int(model_h / stride[i]) length = int(na * h * w) if grid[i].shape[2:4] != (h, w): grid[i] = _make_grid(w, h) outs[row_ind:row_ind + length, 0:2] = (outs[row_ind:row_ind + length, 0:2] * 2. - 0.5 + np.tile( grid[i], (na, 1))) * int(stride[i]) outs[row_ind:row_ind + length, 2:4] = (outs[row_ind:row_ind + length, 2:4] * 2) ** 2 * np.repeat( anchor_grid[i], h * w, axis=0) row_ind += length return outs def post_process_opencv(outputs, model_h, model_w, img_h, img_w, thred_nms, thred_cond): conf = outputs[:, 4].tolist() c_x = outputs[:, 0] / model_w * img_w c_y = outputs[:, 1] / model_h * img_h w = outputs[:, 2] / model_w * img_w h = outputs[:, 3] / model_h * img_h p_cls = outputs[:, 5:] if len(p_cls.shape) == 1: p_cls = np.expand_dims(p_cls, 1) cls_id = np.argmax(p_cls, axis=1) p_x1 = np.expand_dims(c_x - w / 2, -1) p_y1 = np.expand_dims(c_y - h / 2, -1) p_x2 = np.expand_dims(c_x + w / 2, -1) p_y2 = np.expand_dims(c_y + h / 2, -1) areas = np.concatenate((p_x1, p_y1, p_x2, p_y2), axis=-1) areas = areas.tolist() ids = cv2.dnn.NMSBoxes(areas, conf, thred_cond, thred_nms) if len(ids) > 0: return np.array(areas)[ids], np.array(conf)[ids], cls_id[ids] else: return [], [], [] def infer_img(img0, net, model_h, model_w, nl, na, stride, anchor_grid, thred_nms=0.4, thred_cond=0.5): # 图像预处理 img = cv2.resize(img0, (model_w, model_h), interpolation=cv2.INTER_AREA) img = cv2.cvtColor(img, cv2.COLOR_BGR2RGB) img = img.astype(np.float32) / 255.0 blob = np.expand_dims(np.transpose(img, (2, 0, 1)), axis=0) # 模型推理 outs = net.run(None, {net.get_inputs()[0].name: blob})[0].squeeze(axis=0) # 输出坐标矫正 outs = cal_outputs(outs, nl, na, model_w, model_h, anchor_grid, stride) # 检测框计算 img_h, img_w, _ = np.shape(img0) boxes, confs, ids = post_process_opencv(outs, model_h, model_w, img_h, img_w, thred_nms, thred_cond) return boxes, confs, ids if __name__ == "__main__": # 模型加载 model_pb_path = r"/home/eh/Desktop/yolo/daima/best.onnx" so = ort.SessionOptions() net = ort.InferenceSession(model_pb_path, so) # 标签字典 dic_labels = {0: '1', 1: '2', 2: '3', 3: '4', 4: '5', 5: '6', 6: '7', 7: '8', 8: '9' } # 模型参数 model_h = 96 model_w = 96 nl = 3 na = 3 stride = [8., 16., 32.] anchors = [[10, 13, 16, 30, 33, 23], [30, 61, 62, 45, 59, 119], [116, 90, 156, 198, 373, 326]] anchor_grid = np.asarray(anchors, dtype=np.float32).reshape(nl, -1, 2) video = 0 cap = cv2.VideoCapture(video) flag_det = False while True: success, img0 = cap.read() if success: if flag_det: t1 = time.time() det_boxes, scores, ids = infer_img(img0, net, model_h, model_w, nl, na, stride, anchor_grid, thred_nms=0.5, thred_cond=0.5) t2 = time.time() for box, score, id in zip(det_boxes, scores, ids): label = '%s:%.2f' % (dic_labels[id.item()], score) plot_one_box(box.astype(np.int16), img0, color=(255, 0, 0), label=label, line_thickness=None) str_FPS = "FPS: %.2f" % (1. / (t2 - t1)) cv2.putText(img0, str_FPS, (50, 50), cv2.FONT_HERSHEY_COMPLEX, 1, (0, 255, 0), 3) cv2.imshow("video", img0) key = cv2.waitKey(1) & 0xFF if key == ord('q'): break elif key & 0xFF == ord('s'): flag_det = not flag_det print(flag_det) cap.release() cv2.destroyAllWindows() weishabujiance

JonSco
  • 粉丝: 113
上传资源 快速赚钱