Use yolov5 to get workspace_mask.png

[tjc609]

Now I can use the customized workspace_mask.png to obtain the grasping posture of objects within a certain range, but it is not accurate enough. I want to use yolov5 to segment and detect the target first, and obtain the segmentation mask Boolean image of the target. Running the code will Report an error
customized：

use YOLOv5:

Can you help me answer this question? I will be grateful for your help