ContrastiveLosses4VRD
ContrastiveLosses4VRD copied to clipboard
NVIDIA
Implementation for the CVPR2019 paper "Graphical Contrastive Losses for Scene Graph Generation"
Hello, I was wondering if there was a way to run inference on custom image(s) without gt annotations. For example, if I wanted to generate scene-graph using your pre-trained models...
Dear Ji, I ran into this runtime error when trying to evaluate the model with pertained checkpoints: `python ./tools/test_net_rel.py --dataset vg --cfg configs/vg/e2e_faster_rcnn_VGG16_8_epochs_vg_v3_default_node_contrastive_loss_w_so_p_aware_margin_point2_so_weight_point5_no_spt.yaml --load_ckpt trained_models/vg_VGG16/model_step62722.pth --output_dir Outputs/vg_VGG16 --multi-gpu-testing --do_val `...
` ### Adaptively adjust some configs ### cfg.NUM_GPUS = torch.cuda.device_count() original_batch_size = cfg.NUM_GPUS * cfg.TRAIN.IMS_PER_BATCH original_ims_per_batch = cfg.TRAIN.IMS_PER_BATCH original_num_gpus = cfg.NUM_GPUS if args.batch_size is None: args.batch_size = original_batch_size assert (args.batch_size...
When evaluating on VRD dataset，we record different recall@N of k=1, k=10, k=70. But when evaluating on VG dataset, what is the 'k' value to chose ?
I have a question. I don't understand why (in Visual Genome) SGDet gains such a small improvement compared to Neural Motifs whereas SGCls has gains such a larger improvement? Isn't...
How can i run this code with 11G GPU memory machine? Any help is appreciated, thx!!
Could you provide other download pretrained-model links, like baidu disk? Or maybe could you divide the models seperately rather than using a unzipped file ?
I tried to train this model on MSRVTT, but I don`t have a detection.json file, so I want to obtain it by Detectron_pytorch @ 75793bf with the given detection models....
I wonder how you defined the bbox for the subjects and objects in rel_annotations_train.json? Do you have any documentation or code on how you built rel_annotations_train.json? Anyway, I tracked the...
