UNITER
UNITER copied to clipboard
Research code for ECCV 2020 paper "UNITER: UNiversal Image-TExt Representation Learning"
Due to domain conflict, I wonder if you could also release model checkpoints pretrained on out-of-domain data (CC, SBU) only? Thanks a lot!
Hello. I want to pre-train myself. I try running the CMD here: https://github.com/ChenRocks/UNITER#pre-tranining ``` horovodrun -np 8 python pretrain.py --config config/pretrain-indomain-base-8gpu.json \ --output_dir $PRETRAIN_EXP ``` (after launching the docker) But...
After reproducing paper results, I am interested in the model's (ITM) performance on different datasets. To achieve that I have to run the preprocessing pipeline on my own images but...
Hello, I watched that you use two image features (fname_gt, fname) At first, I thought, img_gt features for image masking tasks like MRM, MRFC etc. But I see it is...
Hello, thanks for open-sourcing the amazing work! Do you mind sharing more experiment details for image feature extraction? For example, what are the score threshold, NMS threshold, and the number...
Hi. Why did you decide to use a BCE loss on the ITM pretraining text and a ranking loss on the ITM downstream task? Is there any intuition behind this?...
https://github.com/ChenRocks/UNITER/blob/80d3602d71d65700eab373acb0507e31e251b7e7/data/sampler.py#L41
Hi @ChenRocks , I see the location features for each region is a 7-dimensional vector, so could you please tell me the order of coordinates? Is it [x1, y1, x2,...
Hi all. I'm Henry Lee, a graduate student at Seoul National University. Thanks for sharing your great work for future research. If you don't mind, could you share the fine-tuned...