Missing/wrong annotations in 2D/3D/InstanceSegmentation - Githubissues

SysCV / shift-dev

SHIFT Dataset DevKit - CVPR2022

https://www.vis.xyz/shift

MIT License

101 stars 10 forks source link

Missing/wrong annotations in 2D/3D/InstanceSegmentation #69

Open PierreId opened 2 months ago

PierreId commented 2 months ago

I see many missing and wrong annotations in 2D/3D/InstanceSegmentation. Example in "https://dl.cv.ethz.ch/shift/discrete/images/val/front/det_2d.json" :

name : 00000220_img_front.jpg
videoName: 773a-45da
Only 3 vehicles are present in the annotations, but there are 2 more (parked) vehicles
The 2D box for the police car (vehicle.dodge.charger_police) has very wrong dimensions
In the SemanticSegmentation image, all the info seem correct.

Would it be possible to correct the annotations ?

Thanks