smd
smd copied to clipboard
akarazniewicz
Simple mmdetection CPU inference
Are you planning on adding support for DetectoRS, as the DeformConvs are not supported at all on CPU only inference?
RuntimeError: Error(s) in loading state_dict for FasterRCNN: size mismatch for roi_heads.box_predictor.cls_score.weight: copying a param with shape torch.Size([11, 1024]) from checkpoint, the shape in current model is torch.Size([10, 1024]). size mismatch...
Hi. Is this repo support [Solo ](https://github.com/WXinlong/SOLO)model? I really need to test it on the CPU. Many thanks
Hello your README have a typo on the notebook link: See demo jupyter notebook complete example: https://github.com/akarazniewicz/smd/blob/master/demo/demo.jpnby But the right link is: https://github.com/akarazniewicz/smd/blob/master/demo/demo.ipynb
I used the mask-RCNN R50 model trained in mmdetection v1 in SMD. In the 'results', the boxes, labels and scores can be extracted. However, the mask values are all 0....
@akarazniewicz is it normal to get a very long inference time when using SMD's Mask R-CNN model on an arbitrary CPU? I get between 10s and 30s depending on image...
