update pipeline according to online demo requirements

按在线demo前端的要求，将输出改成单独一个numpy格式的图片
        Link: https://code.alibaba-inc.com/Ali-MaaS/MaaS-lib/codereview/10912907

modelscope/pipelines/cv/image_depth_estimation_pipeline.py

@@ -47,6 +47,9 @@ class ImageDepthEstimationPipeline(Pipeline):
 
     def postprocess(self, inputs: Dict[str, Any]) -> Dict[str, Any]:
         results = self.model.postprocess(inputs)
-        outputs = {OutputKeys.DEPTHS: results[OutputKeys.DEPTHS]}
+        depths = results[OutputKeys.DEPTHS]
+        if isinstance(depths, torch.Tensor):
+            depths = depths.detach().cpu().squeeze().numpy()
+        outputs = {OutputKeys.DEPTHS: depths}
 
         return outputs

tests/pipelines/test_image_depth_estimation.py

@@ -25,7 +25,7 @@ class ImageDepthEstimationTest(unittest.TestCase, DemoCompatibilityCheck):
         estimator = pipeline(Tasks.image_depth_estimation, model=self.model_id)
         result = estimator(input_location)
         depths = result[OutputKeys.DEPTHS]
-        depth_viz = depth_to_color(depths[0].squeeze().cpu().numpy())
+        depth_viz = depth_to_color(depths)
         cv2.imwrite('result.jpg', depth_viz)
 
         print('test_image_depth_estimation DONE')