search

RnDProjectsDeebul / ManojKolpeThesis

Mozilla Public License 2.0
2 stars 1 forks source link

Segmntation dataset with camera poses #12

Open deebuls opened 2 years ago

deebuls commented 2 years ago

An issue to track all dataset with poses and segmentation labels

deebuls commented 2 years ago

https://europe.naverlabs.com/research/computer-vision/proxy-virtual-worlds-vkitti-2/

Please check I think it has both info . Needs to be verified

deebuls commented 2 years ago

@Manojkl were you able to download this dataset ? Does it contiain poses ? Were you able to load the dataset ?

deebuls commented 2 years ago

https://www.a2d2.audi/a2d2/en/tutorial.html

Also seems to be outdoor dataset with extrinsic parameters please check.

Manojkl commented 2 years ago

I am looking at the dataset. I got preliminary results with the Scannet dataset. Thanks for the link.

Manojkl commented 2 years ago

About the Audi a2d2 dataset, it is the same case as with the cityscape dataset. The extrinsic parameters are given with respect to the "frame of references of the car" not with respect to the position of the car in the 3D world. image

Manojkl commented 2 years ago

I downloaded naverlabs dataset; it contains also poses I can load the dataset. However, further processing must be done to prepare the data for training.

Manojkl commented 2 years ago

Found another interesting dataset with intrinsic, extrinsic, and segmentation data. https://github.com/kminoda/VIODE However, the data format is in .bag file format. I am able to extract the IMU data however I am facing a problem with the extraction of image data. By any chance have you worked on it? And anywhere I can find the script to extract the image data out of it?

Manojkl commented 2 years ago

https://www.nuscenes.org/nuscenes?tutorial=maps

Manojkl commented 2 years ago

Pixel accuracy vs IOU https://towardsdatascience.com/metrics-to-evaluate-your-semantic-segmentation-model-6bcb99639aa2