I am trying to run MonoGS on custom dataset. It is not very straight forward, so i have some questions.
I collect dataset using my iphone. I click a video and extract images out from that. If i run the code on mono mode, ie python slam.py --config configs/mono/tum/config_mydataset.yaml. it doesnt run. So i have some questions regarding it.
[ ] How important is camera calibration? if i just add my dataset (rgb images) and leave the calibration file as it is, it might give me wrong results but should it not give some results?
[ ] I did camera calibrations using checkerboard opencv method. But even changing the camera calibrations based on my iphone and my rgb images, it fails. Does the size of the images or encodings matter a lot?
[ ] If i dont want to provide the accelerometer.txt file and groundtruth.txt file, just only the RGBD images or just only the RGB images, is there a way out to do that?
[ ] in the command, python slam.py --config configs/mono/tum/fr1_desk.yaml, why doesnt it work when we dont provide depth images? because in the end, depth images are anyways not being used ?
I am trying to run MonoGS on custom dataset. It is not very straight forward, so i have some questions.
I collect dataset using my iphone. I click a video and extract images out from that. If i run the code on mono mode, ie python slam.py --config configs/mono/tum/config_mydataset.yaml. it doesnt run. So i have some questions regarding it.