def parse_predict(predictions):
"""Interpreting the predictions to score classes and boxes.
Args:
predictions (np.array): [description]contain all the detections info of a frame.
Returns:
[boxes, classes, scores]: the predictions boxes, classes & scores.
"""
label_classes = cfg['labels_list']
bbox_regressions, confs = tf.split(predictions[0], [4, -1], -1)
boxes = decode_bbox_tf(bbox_regressions, priors, cfg['variances'])
##classifications shape :(num_priors,num_classes)
confs = tf.math.softmax(confs, axis=-1)
out_boxes = []
out_labels = []
out_scores = []
for i in range(1, len(label_classes)):
cls_scores = confs[:, i]
score_idx = cls_scores > DETECTION_THRESHOLD
cls_boxes = boxes[score_idx]
cls_scores = cls_scores[score_idx]
nms_idx = compute_nms(cls_boxes, cls_scores, cfg['nms_threshold'],
cfg['max_number_keep'])
cls_boxes = tf.gather(cls_boxes, nms_idx, axis=None)
cls_scores = tf.gather(cls_scores, nms_idx, axis=None)
cls_labels = [i] * cls_boxes.shape[0]
out_boxes.append(cls_boxes)
out_labels.extend(cls_labels)
out_scores.append(cls_scores)
out_boxes = tf.concat(out_boxes, 0)
out_scores = tf.concat(out_scores, 0)
boxes = tf.clip_by_value(out_boxes, 0.0, 1.0).numpy()
classes = np.array(out_labels)
return boxes, classes, out_scores.numpy()
def parse_predict(predictions, priors, cfg):
label_classes = cfg['labels_list']
bbox_regressions, confs = tf.split(predictions[0], [4, -1], axis=-1)
boxes = decode_bbox_tf(bbox_regressions, priors, cfg['variances'])
##classifications shape :(num_priors,num_classes)
confs = tf.math.softmax(confs, axis=-1)
out_boxes = []
out_labels = []
out_scores = []
for c in range(1, len(label_classes)):
cls_scores = confs[:, c]
score_idx = cls_scores > cfg['score_threshold']
cls_boxes = boxes[score_idx]
cls_scores = cls_scores[score_idx]
nms_idx = compute_nms(cls_boxes, cls_scores, cfg['nms_threshold'],
cfg['max_number_keep'])
cls_boxes = tf.gather(cls_boxes, nms_idx)
cls_scores = tf.gather(cls_scores, nms_idx)
cls_labels = [c] * cls_boxes.shape[0]
out_boxes.append(cls_boxes)
out_labels.extend(cls_labels)
out_scores.append(cls_scores)
out_boxes = tf.concat(out_boxes, axis=0)
out_scores = tf.concat(out_scores, axis=0)
boxes = tf.clip_by_value(out_boxes, 0.0, 1.0).numpy()
classes = np.array(out_labels)
scores = out_scores.numpy()
return boxes, classes, scores