remove center constraint in box_crop

fluid/PaddleCV/yolov3/box_utils.py

@@ -138,6 +138,8 @@ def rescale_box_in_input_image(boxes, im_shape, input_size):
     boxes[:, 2] *= fx
     boxes[:, 3] *= fy
     boxes[boxes<0] = 0
+    boxes[:, 2][boxes[:, 2] > (w - 1)] = w - 1
+    boxes[:, 3][boxes[:, 3] > (h - 1)] = h - 1
     return boxes
 
 def box_crop(boxes, labels, scores, crop, img_shape):
@@ -149,15 +151,16 @@ def box_crop(boxes, labels, scores, crop, img_shape):
     boxes[:, 1], boxes[:, 3] = (boxes[:, 1] - boxes[:, 3] / 2) * im_h, (boxes[:, 1] + boxes[:, 3] / 2) * im_h
 
     crop_box = np.array([x, y, x + w, y + h])
-    centers = (boxes[:, :2] + boxes[:, 2:]) / 2.0
-    mask = np.logical_and(crop_box[:2] <= centers, centers <= crop_box[2:]).all(axis=1)
+    # centers = (boxes[:, :2] + boxes[:, 2:]) / 2.0
+    # mask = np.logical_and(crop_box[:2] <= centers, centers <= crop_box[2:]).all(axis=1)
 
     boxes[:, :2] = np.maximum(boxes[:, :2], crop_box[:2])
     boxes[:, 2:] = np.minimum(boxes[:, 2:], crop_box[2:])
     boxes[:, :2] -= crop_box[:2]
     boxes[:, 2:] -= crop_box[:2]
 
-    mask = np.logical_and(mask, (boxes[:, :2] < boxes[:, 2:]).all(axis=1))
+    # mask = np.logical_and(mask, (boxes[:, :2] < boxes[:, 2:]).all(axis=1))
+    mask =  (boxes[:, :2] < boxes[:, 2:]).all(axis=1)
     boxes = boxes * np.expand_dims(mask.astype('float32'), axis=1)
     labels = labels * mask.astype('float32')
     scores = scores * mask.astype('float32')

fluid/PaddleCV/yolov3/image_utils.py

@@ -121,7 +121,7 @@ def random_interp(img, size, interp=None):
     img = cv2.resize(img, None, None, fx=im_scale_x, fy=im_scale_y, interpolation=interp)
     return img
 
-def random_expand(img, gtboxes, max_ratio=2., fill=None, keep_ratio=True, thresh=0.5):
+def random_expand(img, gtboxes, max_ratio=4., fill=None, keep_ratio=True, thresh=0.5):
     if random.random() > thresh:
         return img, gtboxes
 

fluid/PaddleCV/yolov3/learning_rate.py

@@ -23,7 +23,7 @@ from paddle.fluid.layers import control_flow
 
 def exponential_with_warmup_decay(learning_rate, boundaries, values,
                                   warmup_iter, warmup_factor, start_step):
-    global_step = lr_scheduler._decay_step_counter() + start_step
+    global_step = lr_scheduler._decay_step_counter()
 
     lr = fluid.layers.create_global_var(
         shape=[1],

fluid/PaddleCV/yolov3/reader.py

@@ -267,8 +267,8 @@ def train(size=416,
           interval=10,
           pyreader_num=1,
           use_multiprocessing=True,
-          num_workers=8,
-          max_queue=24):
+          num_workers=12,
+          max_queue=32):
     generator = dsr.get_reader('train', size, batch_size, shuffle, mixup_iter, random_sizes)
 
     if not use_multiprocessing:
@@ -302,6 +302,7 @@ def train(size=416,
                 cnt += 1
                 if cnt % intervals == 0:
                     idx = np.random.randint(len(random_sizes))
+                    print("Resizing: ", (idx + 10) * 32)
         finally:
             if enqueuer is not None:
                 enqueuer.stop()

fluid/PaddleCV/yolov3/train.py

@@ -96,7 +96,7 @@ def train():
 
     mixup_iter = cfg.max_iter - cfg.start_iter - cfg.no_mixup_iter
     if cfg.use_pyreader:
-        train_reader = reader.train(input_size, batch_size=int(hyperparams['batch'])/devices_num, shuffle=True, mixup_iter=mixup_iter, random_sizes=random_sizes, interval=10, pyreader_num=devices, use_multiprocessing=cfg.use_multiprocess)
+        train_reader = reader.train(input_size, batch_size=int(hyperparams['batch'])/devices_num, shuffle=True, mixup_iter=mixup_iter, random_sizes=random_sizes, interval=10, pyreader_num=devices_num, use_multiprocessing=cfg.use_multiprocess)
         py_reader = model.py_reader
         py_reader.decorate_paddle_reader(train_reader)
     else: