lsm140

training details

2

Paper says SparseInst is built on Detectron2 [42] and trained over 8 GPUs with a total of 64 images per mini-batch. Q1:It use fp16 or not? Q2: Small or large...

Is mmcv>=2.0.0 and the newest mmdet3.0 compatible with this code?

BPR checkpoint

3

Dear authors, can you provide the BPR model trained on PointRend or SegFix results rather than mask rcnn？

What's the advantage of TensorRT compared with pytorch?Both 18ms. Does this mean that TensorRT has very limited optimization for SOLOV2?

### 📚 The doc issue rtmpose mmpose demo里的image_demo和inference_demo效果区别怎么这么大？如果要实际部署的话，还得加入pose的det模型吗？要双模型？ ### Suggest a potential alternative/fix _No response_

Labeling problems for instance segmentation

2

Anylabeling does not support removing skeletonized areas, so the resulting polygon mask is wrong and not in rle format.

python predict.py后只产生了空的output/predict文件夹，没有预测结果。是哪里需要修改一下吗

feedback

2

Compared to other platforms' assisted segmentation SAMs, labelfree is slower and the segmentation quality is rougher, requiring many clicks to get a good approximation of the segmentation; however, the advantage...