search

Dawn-LX / OpenVoc-VidVRD

Official code for the ICLR2023 paper Compositional Prompt Tuning with Motion Cues for Open-vocabulary Video Relation Detection
37 stars 1 forks source link

About VidOR dataset #7

Open QiueY514 opened 4 months ago

QiueY514 commented 4 months ago

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?

Dawn-LX commented 4 months ago

refer to experiments/RelationCls_VidVRD for the configs for each exp

QiueY514 commented 4 months ago

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.

Dawn-LX commented 4 months ago

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

Dawn-LX commented 3 months ago

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

QiueY514 commented 3 months ago

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!

QiueY514 commented 3 months ago

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!

Dawn-LX commented 2 months ago

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

Dawn-LX commented 2 months ago

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