Feng Li comments

Results 119 comments of


                                            Feng Li

point loss in mask loss

Hi, you can refer to our cited papers for more details about how to sample and its gain.

Transfer learning issue

If your dataset has similar categories as COCO, you can load the corresponding class embedding. But overall class embedding is easy to learn, you can just randomly pick some class...

Training object365 got error "Killing subprocess"

Hey, we did not meet this problem before.

How to train, evaluate and inference on instance segmentation task

Please refer to our paper [Mask DINO](https://arxiv.org/abs/2206.02777).

When will you publish your code?

Happy December! Our code is available now!

AP = 0

Have you ever changed the code? You may clone a clean code and rerun it with our provided command.

error : python inference.py

It seems some tensors are put in CPU will some are in GPU. You can put all tensors on the GPU to fix this problem.

adding DINO's contrastive sampling

Thanks for your efforts. I am busy preparing for the CVPR these days. I notice that you also made modifications to the DN-Deformable-DETR, how about leaving the original DN-Deformable-DETR unchanged...

adding DINO's contrastive sampling

@Vallum Hey, thanks for your efforts, I have added your implemented DINO in the News of ReadMe. May you also provide the running command and checkpoint so others can reproduce...

adding DINO's contrastive sampling

Got it. Can you convert your trained checkpoint weights with 51.0 AP so that the new model can load it? As the main framework remains the same, maybe you just...