Hi, can you provide pretrained model trained on DAVIS2017 only? ('DAVIS only' in Table S5) Thanks.

Hello, Could you update the link for the extended version of the dataset? The current link does not work. Thank you for the amazing work! Hung

Hi, Thanks for such incredible work. I want to train XMem on surgery data. I have videos of robotic surgery (like CholecT80). I see that in the TRAINING.md file, there are four training stages (0: static images; 1: BL30K; 2: longer main training; 3: shorter main training). What would each stage repr...

Hello, I want to ask the training time of the XMem. especially XMem-s03.pth and XMem-s3.pth. I just training the datasets on the DAVIS & Youtube datasets (setting:3). The config is: batchsize 4, GPUs: 4 iterations: 200000, finetune: 20000. It is about 1.5 -2 days. I have printed the training tim...

What is the difference between the XMem.pth and XMem-s012.pth ? Thanks.

I use 8 GPUs in total and set the batch size to 8. But the speed is too slow(slower than use 1 GPUs). What is the problem?

The GUI doesn't appear, and it stuck at this stage. The terminal shows" Got keys from plugin meta data ("webp") QFactoryLoader::QFactoryLoader() checking directory path "/home/xiaoyin/miniconda3/envs/py38/bin/imageformats" ... loaded library "/home/xiaoyin/miniconda3/envs/py38/lib/python3.8/site-pac...

May I ask why you don't compare the FPS of your method with other methods in Table 1 on the Long-time Video dataset?

Hi, thanks for your excellent work! I have some problems about the stages of training.I want to use the pretrained model, like Xmem-s012.pth, to train my own data(frames of a video) for fine-tuning, I set the training stage as 0, but met a error as: RuntimeError: Error(s) in loading state_dict for X...