Can I use it on CPU?

[wangzishuo2002]

Dear Author,
Sorry to bother you, but i truly met a question that I can't solved.
When i run webcam.py on MacOS(only has cpu, and I have trained the weight on other server), it shows as below:

/Users/wangzishuo/FCOS-master/bin/python /Users/wangzishuo/Desktop/FCOS-master/demo/webcam.py
/Users/wangzishuo/FCOS-master/lib/python3.7/site-packages/torch/functional.py:568: UserWarning: torch.meshgrid: in an upcoming release, it will be required to pass the indexing argument. (Triggered internally at /Users/distiller/project/pytorch/aten/src/ATen/native/TensorShape.cpp:2228.)
return _VF.meshgrid(tensors, **kwargs) # type: ignore[attr-defined]
Traceback (most recent call last):
File "/Users/wangzishuo/Desktop/FCOS-master/demo/webcam.py", line 80, in
main()
File "/Users/wangzishuo/Desktop/FCOS-master/demo/webcam.py", line 71, in main
composite = coco_demo.run_on_opencv_image(img)
File "/Users/wangzishuo/Desktop/FCOS-master/demo/predictor.py", line 93, in run_on_opencv_image
predictions = self.compute_prediction(image)
File "/Users/wangzishuo/Desktop/FCOS-master/demo/predictor.py", line 126, in compute_prediction
predictions = self.model(image_list)
File "/Users/wangzishuo/FCOS-master/lib/python3.7/site-packages/torch/nn/modules/module.py", line 1110, in _call_impl
return forward_call(*input, **kwargs)
File "/Users/wangzishuo/Desktop/FCOS-master/fcos_core/modeling/detector/generalized_rcnn.py", line 50, in forward
proposals, proposal_losses = self.rpn(images, features, targets)
File "/Users/wangzishuo/FCOS-master/lib/python3.7/site-packages/torch/nn/modules/module.py", line 1110, in _call_impl
return forward_call(*input, **kwargs)
File "/Users/wangzishuo/Desktop/FCOS-master/fcos_core/modeling/rpn/fcos/fcos.py", line 164, in forward
centerness, images.image_sizes
File "/Users/wangzishuo/Desktop/FCOS-master/fcos_core/modeling/rpn/fcos/fcos.py", line 181, in _forward_test
centerness, image_sizes
File "/Users/wangzishuo/FCOS-master/lib/python3.7/site-packages/torch/nn/modules/module.py", line 1110, in _call_impl
return forward_call(*input, **kwargs)
File "/Users/wangzishuo/Desktop/FCOS-master/fcos_core/modeling/rpn/fcos/inference.py", line 137, in forward
boxlists = self.select_over_all_levels(boxlists)
File "/Users/wangzishuo/Desktop/FCOS-master/fcos_core/modeling/rpn/fcos/inference.py", line 150, in select_over_all_levels
result = boxlist_ml_nms(boxlists[i], self.nms_thresh)
File "/Users/wangzishuo/Desktop/FCOS-master/fcos_core/structures/boxlist_ops.py", line 55, in boxlist_ml_nms
keep = _box_ml_nms(boxes, scores, labels.float(), nms_thresh)
RuntimeError: CPU version not implemented

Could you please tell me how to use it correctly on cpu? Thanks a lot!
maybe has some problem with fcos_core/csrc/ml_nms.h, but it's too complicated and i don't know how to change the c++ code to let it run on cpu.