def _to_original_scale(boxes):
minmax_boxes = to_minmax(boxes)
minmax_boxes[:, 0] *= self.output_width
minmax_boxes[:, 2] *= self.output_width
minmax_boxes[:, 1] *= self.output_height
minmax_boxes[:, 3] *= self.output_height
return minmax_boxes.astype(np.int)
def _to_original_scale(boxes):
minmax_boxes = to_minmax(boxes)
minmax_boxes[:, 0] *= 224
minmax_boxes[:, 2] *= 224
minmax_boxes[:, 1] *= 224
minmax_boxes[:, 3] *= 224
return minmax_boxes.astype(np.int)
def detect(self, image, anchors, net_size=416):
image_h, image_w, _ = image.shape
new_image = preprocess_input(image, net_size)
# 3. predict
yolos = self.predict(new_image)
boxes_ = postprocess_ouput(yolos, anchors, net_size, image_h, image_w)
if len(boxes_) > 0:
boxes, probs = boxes_to_array(boxes_)
boxes = to_minmax(boxes)
labels = np.array([b.get_label() for b in boxes_])
else:
boxes, labels, probs = [], [], []
return boxes, labels, probs