CODEWITHSUNDEEP
3dcomputer-visionlabelkitti
kepdsadef
kepdsadef

Reputation: 11

Labeling Data - Stereo Vision - KITTI Dataset

I am trying to understand going from image labeling 2D to 3D stereo vision.

When you label 2D images each image has bounding boxes that you can output as a txt file which gives the location and size of the object, what I don't understand is when you have a pair of left and right images, do you label both images, or the disparity map, or none of the above?

For example the KITTI dataset contains a txt file with all the bounding box information, but only one txt file is given, which image would they have labelled the left or right?

Ultimately I want to be doing 3D object detection and not 2D but I am stuck trying to understand how to go about labeling a pair..

Thank you if you can provide any help in understanding.

I have tried labelling in 2D but don't understand how to then label a stereo pair of images left and right compared to traditional labelling of one image in 2D

Upvotes: 0

Views: 76

Answers (0)

Related Questions