if cv2.waitKey(1) & 0xFF == ord('q'): break camera.release() cv2.destroyAllWindows() if __name__ == "__main__": from models.ssd import SSD300 from utils.boxes import create_prior_boxes from utils.boxes import to_point_form dataset_name = 'VOC2007' # weights_path = '../trained_models/SSD300_weights.hdf5' # weights_path = '../trained_models/weights.07-3.59.hdf5' # weights_path = '../trained_models/weights.03-3.37.hdf5' # weights_path = '../trained_models/weights.150-3.57.hdf5' # weights_path = '../trained_models/weights.12-4.20.hdf5' # weights_path = '../trained_models/weights.02-3.44.hdf5' # weights_path = '../trained_models/weights.22-5.01.hdf5' # weights_path = '../trained_models/weights.79-6.66.hdf5' # weights_path = '../trained_models/weights.64-6.52.hdf5' # weights_path = '../trained_models/weights.22-3.85.hdf5' # weights_path = '../trained_models/weights.50-3.92.hdf5' weights_path = '../trained_models/weights.04-3.79.hdf5' model = SSD300(weights_path=weights_path) prior_boxes = to_point_form(create_prior_boxes()) # prior_boxes = to_point_form(prior_boxes) video = VideoDemo(prior_boxes, dataset_name) video.start_video(model)
# Preprocess precision to be a non-decreasing array for i in range(len(precision) - 2, -1, -1): precision[i] = np.maximum(precision[i], precision[i + 1]) indices = np.where(recall[1:] != recall[:-1])[0] + 1 average_precision = np.sum( (recall[indices] - recall[indices - 1]) * precision[indices]) return average_precision dataset_name = 'VOC2007' data_prefix = '../datasets/VOCtest/VOCdevkit/VOC2007/Annotations/' image_prefix = '../datasets/VOCtest/VOCdevkit/VOC2007/JPEGImages/' weights_path = '../trained_models/weights_SSD300.hdf5' model = SSD300(weights_path=weights_path) prior_boxes = create_prior_boxes(model) input_shape = model.input_shape[1:3] class_threshold = .1 iou_threshold = .5 average_precisions = [] for ground_truth_class_arg in range(1, 21): labels = [] scores = [] class_names = get_class_names(dataset_name) #ground_truth_class_arg = class_arg selected_classes = [class_names[0]] + [class_names[ground_truth_class_arg]] num_ground_truth_boxes = 0 class_decoder = get_arg_to_class(class_names) num_classes = len(class_names)
from preprocessing import get_image_size from models import SSD300 from metrics import compute_average_precision from metrics import compute_precision_and_recall from utils.boxes import create_prior_boxes from utils.boxes import calculate_intersection_over_union from utils.boxes import denormalize_boxes from utils.inference import infer_from_path dataset_name = 'VOC2007' image_prefix = '../datasets/VOCdevkit/VOC2007/JPEGImages/' weights_path = '../trained_models/SSD300_weights.hdf5' model = SSD300(weights_path=weights_path) prior_boxes = create_prior_boxes() input_shape = model.input_shape[1:3] class_threshold = .1 iou_nms_threshold = .45 iou_threshold = .5 num_classes = 21 image_prefix = '../datasets/VOCdevkit/VOC2007/JPEGImages/' with_difficult_objects = False split = 'test' class_names = get_class_names(dataset_name) class_names = class_names[1:] average_precisions = [] for class_name in class_names: selected_classes = ['background'] + [class_name]
def _main(args): start_time = timer() input_path = os.path.expanduser(args.input_path) output_path = os.path.expanduser(args.output_path) if not os.path.exists(output_path): print('Creating output path {}'.format(output_path)) os.mkdir(output_path) logging.basicConfig(filename=output_path + "/tracking.log", level=logging.DEBUG) #parse car positions and angles print("Parsing timestamps and oxts files...") if args.oxts.startswith('..'): parse_oxts(input_path + "/" + args.oxts, input_path + "/" + args.time_stamps) else: parse_oxts(args.oxts, args.time_stamps) print("Done. Data acquired.") dataset_name = 'VOC2007' NUM_CLASSES = 21 weights_filename = args.model_path model = SSD300(num_classes=NUM_CLASSES) prior_boxes = create_prior_boxes(model) model.load_weights(weights_filename) # drawing stuff # Generate colors for drawing bounding boxes. hsv_tuples = [(x / NUM_CLASSES, 1., 1.) for x in range(NUM_CLASSES)] colors = list(map(lambda x: colorsys.hsv_to_rgb(*x), hsv_tuples)) colors = list( map(lambda x: (int(x[0] * 255), int(x[1] * 255), int(x[2] * 255)), colors)) random.seed(10101) # Fixed seed for consistent colors across runs. random.shuffle(colors) # Shuffle colors to decorrelate adjacent classes. random.seed(None) # Reset seed to default. class_names = get_class_names(dataset_name) box_scale_factors=[.1, .1, .2, .2] # maybe adjust to our image size background_index=0 lower_probability_threshold=.1 iou_threshold=.2 num_classes = len(class_names) # sould be equal to NUM_CLASSES arg_to_class = dict(zip(list(range(num_classes)), class_names)) frame_idx = 0 for image_file in os.listdir(input_path): try: image_type = imghdr.what(os.path.join(input_path, image_file)) if not image_type: print("frame dropped") continue except IsADirectoryError: print("frame dropped") continue image = Image.open(os.path.join(input_path, image_file)) image_data = np.array(image) selected_boxes = predict(model, image_data, prior_boxes, image_data.shape[0:2], num_classes, lower_probability_threshold, iou_threshold, background_index, box_scale_factors) if selected_boxes is not None: x_mins = selected_boxes[:, 0] y_mins = selected_boxes[:, 1] x_maxs = selected_boxes[:, 2] y_maxs = selected_boxes[:, 3] classes = selected_boxes[:, 4:] num_boxes = len(selected_boxes) else: num_boxes = 0 print("frame dropped, no boxes") print('Found {} boxes for {}'.format(num_boxes, image_file)) font = ImageFont.truetype( font='font/FiraMono-Medium.otf', size=11) # np.floor(3e-2 * image.size[1] + 0.5).astype('int32') thickness = (image_data.shape[0] + image_data.shape[1]) // 300 logging.info("Img: " + str(image_file)) boxes_data = [] for i in range(num_boxes): xmin = int(x_mins[i]) ymin = int(y_mins[i]) xmax = int(x_maxs[i]) ymax = int(y_maxs[i]) box_class_scores = classes[i] label_class = np.argmax(box_class_scores) score = box_class_scores[label_class] predicted_class = arg_to_class[label_class] box = [ymin, xmin, ymax, xmax] box = [max(0,v) for v in box] # sometimes it's negative. # log positions obj_coord = np.array([]) if predicted_class in ["person", "bicycle", "car", "motorbike", "bus", "train", "truck"] and score > 0.2: #object and classes to track if predicted_class in ["bus", "bruck"]: #vehicle predicted_class = "car" obj_coord = computeCoordinates(box, frame_idx) if obj_coord is not None: hist = histogram(image_data, box) #create data and store it boxes_data.append({ 'predicted_class': predicted_class, 'score': float(score), 'coord': obj_coord, 'hist': hist }) logging.info(predicted_class + " :" + str(obj_coord) + " | " + str(np.linalg.norm(obj_coord))) # end log positions if saveImages: if obj_coord is not None: label = '{} {:.2f} {} {:.2f}'.format(predicted_class, score, str(obj_coord), np.linalg.norm(obj_coord)) else: label = '{} {:.2f}'.format(predicted_class, score) draw = ImageDraw.Draw(image) label_size = draw.textsize(label, font) top, left, bottom, right = box top = max(0, np.floor(top + 0.5).astype('int32')) left = max(0, np.floor(left + 0.5).astype('int32')) bottom = min(image.size[1], np.floor(bottom + 0.5).astype('int32')) right = min(image.size[0], np.floor(right + 0.5).astype('int32')) print(label, (left, top), (right, bottom)) if top - label_size[1] >= 0: text_origin = np.array([left, top - label_size[1]]) else: text_origin = np.array([left, top + 1]) for i in range(thickness): draw.rectangle( [left + i, top + i, right - i, bottom - i], outline=colors[label_class]) draw.rectangle( [tuple(text_origin), tuple(text_origin + label_size)], fill=colors[label_class]) draw.text(text_origin, label, fill=(0, 0, 0), font=font) del draw frame_idx += 1 global data_frames data_frames.append(boxes_data) if saveImages: image.save(os.path.join(output_path, image_file), quality=80) now = timer() start_trj_time = timer() print("Time elapsed CNN: " + str(now - start_time) + " seconds") print("Calculating trajectories...") calculate_trajectories() now = timer() print("Done. Time elapsed: " + str(now - start_trj_time) + " seconds\n\n") print("Total time elapsed: " + str(now - start_time) + " seconds")