search

MoonBlvd / Detection-of-Traffic-Anomaly

This is the repo for our Detection of Traffic Anomaly (DoTA) dataset.
MIT License
200 stars 38 forks source link

Extracted data(Dota_fol_train_dota) bbox>1280 #27

Open Ryan7180 opened 1 year ago

Ryan7180 commented 1 year ago

Hi brain,brilliant work!I read extracted data you support for FOL from DOTA(https://drive.google.com/drive/folders/1IVCedrlPg03Fsg4tqDA2cWYlcdrsKUsp?usp=sharing).When I read the file "0qfbmt4G8Rw_000435.pkl",I find that the bbox>1280,is the image size = 1280 X 720? Why the bbox size larger than the image?Can you explain it,Thx. 2022-11-05 23-04-35屏幕截图

MoonBlvd commented 1 year ago

Yes the solution is 1280*720. I think these tracked bounding boxes are from object detection and object tracking output so either of these two modules might have decided to predict the box vertices over the width of image.

Ryan7180 commented 1 year ago

Yes the solution is 1280*720. I think these tracked bounding boxes are from object detection and object tracking output so either of these two modules might have decided to predict the box vertices over the width of image.

Do you mean this can happen under normal circumstances and I can still use these data?

handsomexd-ai commented 5 months ago

Yes the solution is 1280*720. I think these tracked bounding boxes are from object detection and object tracking output so either of these two modules might have decided to predict the box vertices over the width of image.

Do you mean this can happen under normal circumstances and I can still use these data?

Can you send me the content of this download link? This link is no longer valid.My email:3020244106@tju.edu.cn

Ryan7180 commented 4 months ago

This may be what you want. Good luck.

链接：https://pan.baidu.com/s/11xKKFM20u7UfDc3YWUGxrA?pwd=1ru6 提取码：1ru6

On Tue, May 7, 2024 at 9:38 AM Xiao Dong @.***> wrote:

Yes the solution is 1280*720. I think these tracked bounding boxes are from object detection and object tracking output so either of these two modules might have decided to predict the box vertices over the width of image.

Do you mean this can happen under normal circumstances and I can still use these data?

Can you send me the content of this download link? This link is no longer valid.My @.***

— Reply to this email directly, view it on GitHub https://github.com/MoonBlvd/Detection-of-Traffic-Anomaly/issues/27#issuecomment-2097207364, or unsubscribe https://github.com/notifications/unsubscribe-auth/AYYQCGGGSVXB7MNDAB3UFTDZBAWAJAVCNFSM6AAAAAARX63O2KVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDAOJXGIYDOMZWGQ . You are receiving this because you authored the thread.Message ID: @.***>

shyern commented 3 months ago

This may be what you want. Good luck. 链接：https://pan.baidu.com/s/11xKKFM20u7UfDc3YWUGxrA?pwd=1ru6 提取码：1ru6 … On Tue, May 7, 2024 at 9:38 AM Xiao Dong @.> wrote: Yes the solution is 1280720. I think these tracked bounding boxes are from object detection and object tracking output so either of these two modules might have decided to predict the box vertices over the width of image. Do you mean this can happen under normal circumstances and I can still use these data? Can you send me the content of this download link? This link is no longer valid.My **@. — Reply to this email directly, view it on GitHub <#27 (comment)>, or unsubscribe https://github.com/notifications/unsubscribe-auth/AYYQCGGGSVXB7MNDAB3UFTDZBAWAJAVCNFSM6AAAAAARX63O2KVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDAOJXGIYDOMZWGQ . You are receiving this because you authored the thread.Message ID: @.>

Hi, have you extracted the ego motion of DoTA? could you share them? (my email: hshi26@uic.edu) I have tried to extract ego motion using ORB_SLAM2, but the results are very bad.