[to #42322933] fix: remove restriction on image resolution in the preprocessing stage

去掉预处理阶段resize 到720x1280的限制，支持输出与原始输入图像分辨率对应的结果

Link: https://code.alibaba-inc.com/Ali-MaaS/MaaS-lib/codereview/11762135

modelscope/models/cv/image_driving_perception/image_driving_percetion_model.py

@@ -37,6 +37,7 @@ class YOLOPv2(TorchModel):
             [pred, anchor_grid], seg, ll = self.model(img)
         return {
             'img_hw': data['img'].shape[2:],
+            'ori_img_shape': data['ori_img_shape'],
             'pred': pred,
             'anchor_grid': anchor_grid,
             'driving_area_mask': seg,

modelscope/models/cv/image_driving_perception/preprocessor.py

@@ -102,7 +102,7 @@ class ImageDrivingPerceptionPreprocessor(Preprocessor):
             img = self._check_image(img)
         else:
             raise Exception('img is None')
-        img = cv2.resize(img, output_shape, interpolation=cv2.INTER_LINEAR)
+        ori_h, ori_w, _ = img.shape
         img = self._letterbox(img, new_shape)[0]
         img = img.transpose(2, 0, 1)  # to 3x640x640
 
@@ -117,4 +117,5 @@ class ImageDrivingPerceptionPreprocessor(Preprocessor):
 
         return {
             'img': img,
+            'ori_img_shape': (ori_h, ori_w),
         }

modelscope/models/cv/image_driving_perception/utils.py

@@ -29,10 +29,7 @@ def split_for_trace_model(pred=None, anchor_grid=None):
     return pred
 
 
-def scale_coords(img1_shape,
-                 coords,
-                 img0_shape=(720, 1280, 3),
-                 ratio_pad=None):
+def scale_coords(img1_shape, coords, img0_shape=(720, 1280), ratio_pad=None):
     # Rescale coords (xyxy) from img1_shape to img0_shape
     if ratio_pad is None:  # calculate from img0_shape
         gain = min(img1_shape[0] / img0_shape[0],
@@ -190,19 +187,17 @@ def box_iou(box1, box2):
                     )  # iou = inter / (area1 + area2 - inter)
 
 
-def driving_area_mask(seg=None):
-    da_predict = seg[:, :, 12:372, :]
+def driving_area_mask(da_predict=None, out_shape=(720, 1280)):
     da_seg_mask = torch.nn.functional.interpolate(
-        da_predict, scale_factor=2, mode='bilinear')
+        da_predict, size=out_shape, mode='bilinear')
     _, da_seg_mask = torch.max(da_seg_mask, 1)
     da_seg_mask = da_seg_mask.int().squeeze().cpu().numpy()
     return da_seg_mask
 
 
-def lane_line_mask(ll=None):
-    ll_predict = ll[:, :, 12:372, :]
+def lane_line_mask(ll_predict=None, out_shape=(720, 1280)):
     ll_seg_mask = torch.nn.functional.interpolate(
-        ll_predict, scale_factor=2, mode='bilinear')
+        ll_predict, size=out_shape, mode='bilinear')
     ll_seg_mask = torch.round(ll_seg_mask).squeeze(1)
     ll_seg_mask = ll_seg_mask.int().squeeze().cpu().numpy()
     return ll_seg_mask

modelscope/pipelines/cv/image_driving_perception_pipeline.py

@@ -84,13 +84,17 @@ class ImageDrivingPerceptionPipeline(Pipeline):
         # Apply NMS
         pred = non_max_suppression(pred)
 
-        da_seg_mask = driving_area_mask(inputs['driving_area_mask'])
-        ll_seg_mask = lane_line_mask(inputs['lane_line_mask'])
+        h, w = inputs['ori_img_shape']
+        da_seg_mask = driving_area_mask(
+            inputs['driving_area_mask'], out_shape=(h, w))
+        ll_seg_mask = lane_line_mask(
+            inputs['lane_line_mask'], out_shape=(h, w))
 
         for det in pred:  # detections per image
             if len(det):
-                # Rescale boxes from img_size to (720, 1280)
-                det[:, :4] = scale_coords(inputs['img_hw'], det[:, :4]).round()
+                # Rescale boxes from img_size to (h, w)
+                det[:, :4] = scale_coords(inputs['img_hw'], det[:, :4],
+                                          (h, w)).round()
 
         results_dict[OutputKeys.BOXES] = det[:, :4].cpu().numpy()
         results_dict[OutputKeys.MASKS].append(da_seg_mask)

modelscope/utils/cv/image_utils.py

@@ -498,8 +498,6 @@ def show_image_driving_perception_result(img,
                                          results,
                                          out_file='result.jpg',
                                          if_draw=[1, 1, 1]):
-    assert img.shape == (720, 1280,
-                         3), 'input image shape need fix to (720, 1280, 3)'
     bboxes = results.get(OutputKeys.BOXES)
     if if_draw[0]:
         for x in bboxes: