Mingqiao Ye

This looks like a data path issue. You can check if the image corresponding to the path `/content/drive/MyDrive/Iris-and-Needle-Segmentation-3/train/images/SID0615_jpg.rf.8dd4aeb70ce910df9c8716e3af21b2cd.jpg` is still there. Or is your google drive disconnected?

Hi, the evaluation script in the train folder is `python -m torch.distributed.launch --nproc_per_node=1 train.py --checkpoint ./pretrained_checkpoint/sam_vit_l_0b3195.pth --model-type vit_l --output work_dirs/hq_sam_l --eval --restore-model work_dirs/hq_sam_l/epoch_11.pth` In this script, we load sam_vit_l_0b3195.pth for...

Hi, when evaluating the four HQ datasets, we only need to load the mask_decoder of hq_sam. We do it this way because it saves storage space during training. For example,...

Hi, we have updated the [script](https://github.com/SysCV/sam-hq#onnx-export) and [example colab notebook](https://colab.research.google.com/drive/11U2La49c2IxahzJkAV-EzPqEH3cz_5hq?usp=sharing) for exporting and using ONNX model.

> > Hi, we have updated the [script](https://github.com/SysCV/sam-hq#onnx-export) and [example colab notebook](https://colab.research.google.com/drive/11U2La49c2IxahzJkAV-EzPqEH3cz_5hq?usp=sharing) for exporting and using ONNX model. > > Thank you for your nice work! But, how can we...

Hi, I would like to suggest you add some negative points if there are some background regions in the mask output. You can sample some points in the unwanted area...

Hi, thanks for watching HQ-SAM. This is a model weight mismatch problem. As is described in [Getting Started](https://github.com/SysCV/sam-hq#getting-started), to load sam_hq_vit_l.pth, we need to use [demo_hqsam.py](https://github.com/SysCV/sam-hq/blob/main/demo/demo_hqsam.py) instead of [demo_sam.py](https://github.com/SysCV/sam-hq/blob/main/demo/demo_sam.py). The...

Hi, for a multi-object dataset, you could simply modify the dataloader to sample one mask each time randomly. Or you could load all masks and use them as prompts. Then...

Hi, thanks for watching our work! Yes, we have two pre-trained checkpoints for the initialization of training. And they are merged at the end of training in [this line](https://github.com/SysCV/sam-hq/blob/main/train/train.py#L493). The...

It is trained on SA-1B dataset proposed in [SAM](https://github.com/facebookresearch/segment-anything).