MMSceneGraph
MMSceneGraph copied to clipboard
Kenneth-Wong
ICCV 2021: A brand new hub for Scene Graph Generation methods based on MMdetection (2021). The pipeline of from detection, scene graph generation to downstream tasks (e.g., image cpationing) is suppor...
Hello, this is an interesting project, but I did not find your pre-trained model when I was reproducing your program. Can you provide us with the model "./experiments/VG_COCOremap_MASKTRANS_mask_rcnn_x101_64x4d_fpn_1x/latest.pth" ?
Models can not be downloaded from either from s3 or aliyun. Please provide working URLs. Regards
**Describe the Issue** Train FasterRCNN on VG dataset. Error code: TypeError: conv2d(): argument 'input' (position 1) must be Tensor, not DataContainer **Reproduction** 1. What command, code, or script did you...
When I run "python tools/train.py configs/scene_graph/VG_SgDet_heth_area_mask_X_rcnn_x101_64x4d_fpn_1x.py", There was an error: File "/home/XXX/MMSceneGraph/mmdet/models/relation_heads/het_head.py", line 117, in forward roi_feats, union_feats, det_result = self.frontend_features(img, img_meta, det_result, gt_result) File "/home/XXX/MMSceneGraph/mmdet/models/relation_heads/relation_head.py", line 229, in frontend_features...
Hi, can you provide the 'coco_train_input.pkl', 'coco_test_input.pkl' and other used files in the caption task.
Metadata
Owner
Metadata
ICCV 2021: A brand new hub for Scene Graph Generation methods based on MMdetection (2021). The pipeline of from detection, scene graph generation to downstream tasks (e.g., image cpationing) is suppor...