About VidOR dataset

Question

About VidOR dataset

Opened this issue 7 months ago · 8 comments

Thank you for your great work! How can I obtain the file pre-prepared_data/vidor_ObjectTextEmbeddings_v2.pth? And are you planning to release the relation classfication experiment config code?

Answer 1 · 2024-03-06T08:33:37.000Z

refer to experiments/RelationCls_VidVRD for the configs for each exp

Answer 2 · 2024-03-07T08:05:44.000Z

refer to experiments/RelationCls_VidVRD for the configs for each exp

Thanks! And how can I obtain the file pre-prepared_data/vidor_ObjectTextEmbeddings_v2.pth for model training? Additionally, why are the released VidOR tracking results at the video level? In reference to the VidSGG-TrajDatPrepare project, the required tracking result files are at the segment level. Could you consider releasing the segment-level tracking results for the VidOR dataset as well? I would be very grateful if you could reply.

Answer 3 · 2024-03-11T02:38:49.000Z

refer to experiments/RelationCls_VidVRD for the configs for each exp

Thanks! And how can I obtain the file pre-prepared_data/vidor_ObjectTextEmbeddings_v2.pth for model training? Additionally, why are the released VidOR tracking results at the video level? In reference to the VidSGG-TrajDatPrepare project, the required tracking result files are at the segment level. Could you consider releasing the segment-level tracking results for the VidOR dataset as well? I would be very grateful if you could reply.

sorry for the late replay. I will check this in our lab's server, and update it to this repo tonight.

About the tracking results, I will figure out this later and reply you tomorrow

Answer 4 · 2024-03-11T14:12:26.000Z

refer to experiments/RelationCls_VidVRD for the configs for each exp

Thanks! And how can I obtain the file pre-prepared_data/vidor_ObjectTextEmbeddings_v2.pth for model training? Additionally, why are the released VidOR tracking results at the video level? In reference to the VidSGG-TrajDatPrepare project, the required tracking result files are at the segment level. Could you consider releasing the segment-level tracking results for the VidOR dataset as well? I would be very grateful if you could reply.

hi, I have uploaded vidor_ObjectTextEmbeddings_v2: refer to https://github.com/Dawn-LX/OpenVoc-VidVRD/blob/master/vidor_ObjectTextEmbeddings_v2.pth

Answer 5 · 2024-03-12T06:02:59.000Z

refer to experiments/RelationCls_VidVRD for the configs for each exp

Thanks! And how can I obtain the file pre-prepared_data/vidor_ObjectTextEmbeddings_v2.pth for model training? Additionally, why are the released VidOR tracking results at the video level? In reference to the VidSGG-TrajDatPrepare project, the required tracking result files are at the segment level. Could you consider releasing the segment-level tracking results for the VidOR dataset as well? I would be very grateful if you could reply.

hi, I have uploaded vidor_ObjectTextEmbeddings_v2: refer to https://github.com/Dawn-LX/OpenVoc-VidVRD/blob/master/vidor_ObjectTextEmbeddings_v2.pth

Thank you for your generous sharing!

Answer 6 · 2024-03-22T02:20:07.000Z

refer to experiments/RelationCls_VidVRD for the configs for each exp

Thanks! And how can I obtain the file pre-prepared_data/vidor_ObjectTextEmbeddings_v2.pth for model training? Additionally, why are the released VidOR tracking results at the video level? In reference to the VidSGG-TrajDatPrepare project, the required tracking result files are at the segment level. Could you consider releasing the segment-level tracking results for the VidOR dataset as well? I would be very grateful if you could reply.

sorry for the late replay. I will check this in our lab's server, and update it to this repo tonight.

About the tracking results, I will figure out this later and reply you tomorrow

Hi! I find another problem. The released VidOR trajectory tracking results is VidORtrainVideoLevel_tracking_results_gt_th-15-5.zip, while the tracking result dir in OpenVoc-VidVRD/experiments/TrajCls_VidOR/NoBgEmb/cfg_.py is VidORtrainVideoLevel_tracking_results_th-15-5. So do you plan to release the detected video level tracking results, due to I only can download the gt video level tracking results? Thank you!

Answer 7 · 2024-04-17T07:17:28.000Z

refer to experiments/RelationCls_VidVRD for the configs for each exp

Thanks! And how can I obtain the file pre-prepared_data/vidor_ObjectTextEmbeddings_v2.pth for model training? Additionally, why are the released VidOR tracking results at the video level? In reference to the VidSGG-TrajDatPrepare project, the required tracking result files are at the segment level. Could you consider releasing the segment-level tracking results for the VidOR dataset as well? I would be very grateful if you could reply.

sorry for the late replay. I will check this in our lab's server, and update it to this repo tonight.
About the tracking results, I will figure out this later and reply you tomorrow

Hi! I find another problem. The released VidOR trajectory tracking results is VidORtrainVideoLevel_tracking_results_gt_th-15-5.zip, while the tracking result dir in OpenVoc-VidVRD/experiments/TrajCls_VidOR/NoBgEmb/cfg_.py is VidORtrainVideoLevel_tracking_results_th-15-5. So do you plan to release the detected video level tracking results, due to I only can download the gt video level tracking results? Thank you!

VidORtrainVideoLevel_tracking_results_th-15-5 is 9 GB, and I am uploading it to the MEGA cloud.
by the way, I am also uploading VidOR_TrajFeatures256_th-15-5.zip to MEGA cloud, which is 12 GB

Answer 8 · 2024-04-26T02:17:08.000Z

Hi! I find another problem. The released VidOR trajectory tracking results is VidORtrainVideoLevel_tracking_results_gt_th-15-5.zip, while the tracking result dir in OpenVoc-VidVRD/experiments/TrajCls_VidOR/NoBgEmb/cfg_.py is VidORtrainVideoLevel_tracking_results_th-15-5. So do you plan to release the detected video level tracking results, due to I only can download the gt video level tracking results? Thank you!

VidORtrainVideoLevel_tracking_results_th-15-5 is 9 GB, and I am uploading it to the MEGA cloud. by the way, I am also uploading VidOR_TrajFeatures256_th-15-5.zip to MEGA cloud, which is 12 GB

hello, now, all VidOR-traj data is complete: you can find them at (MEGA cloud link).

refer to https://github.com/Dawn-LX/OpenVoc-VidVRD/tree/master?tab=readme-ov-file#vidor for details