VinVL
VinVL copied to clipboard
pzzhang
project page for VinVL
It seems that [this](https://penzhanwu2.blob.core.windows.net/sgg/sgg_benchmark/vinvl_model_zoo/vinvl_vg_x152c4.pth) is not working now. Can you explain how to download the pretrained model?
Hi, is VG dataset pretrained image features included in COCO dataset ？
Hi @pzzhang Please, how to run your source code in repo ? Thanks
Thanks for taking the time to release the pre-trained models and extracted features for all datasets you've used. However, I noticed that the `COCO 2014 Train/Val Image Features (~50G)` doesn't...
Is there an exact list of image IDs that were used in training and tuning of hyperparameters (ie images that were part of any dev sets that were potentially used)?...
Hi, wonderful project! Here I have a question the visualization. The command for visualizing the detections from the pretrained models in your README.md is: `python tools/demo/demo_image.py --config_file sgg_configs/vgattr/vinvl_x152c4.yaml --img_file ../maskrcnn-benchmark-1/datasets1/imgs/woman_fish.jpg...
In https://github.com/pzzhang/VinVL/blob/main/DOWNLOAD.md#pre-trained-models you explain that you merged COCO with stuff, Visual Genome, Objects365 and Open Images into one dataset. Could you please provide this merged dataset or scripts on how...
Hi! I have a question about the 1K testing set of image-text retrieval. In your dataset. There is a file "test_img_keys_1k.tsv". Do you test your model on this 1K testing...
Hi, Is it possible to change the region feature size from 2048 to 500 for finetuning captioning model? If yes, what should I change? Alternatively is it ok to use...
