AdaBins VS Depth-Anything

if os.path.exists("AdaBins") is not True: gitclone("https://github.com/shariqfarooq123/AdaBins.git")

AdaBins: Depth Estimation using Adaptive Bins * https://github.com/shariqfarooq123/AdaBins * https://arxiv.org/abs/2011.14141

I was curious what would happen if I ran a few of these models on the same input and calculated the average. So, I ran (1) [AdaBins](https://github.com/shariqfarooq123/AdaBins) (NYU + KITTI models), (2) [DiverseDepth](https://github.com/YvanYin/DiverseDepth), (3) [MiDaS](https://github.com/intel-isl/MiDaS), and (4) [SGDepth](https://github.com/ifnspaml/SGDepth), and calculated a weighted-average depth prediction.

Depth estimation improved a lot as well e.g. with Depth-Anything [0]. But those are mostly relative depth instead of metric. Also when even converted to metric they still seems have a lot of pointclouds at the edges that have to be pruned - visible in this blog [1]. Looks like those models trained on Lidar or Stereo depthmaps that has this limitations. I think we don't have enough clean training data for 3d unless we maybe train on synthetic data (then we can have plenty, generate realistic scene in Unreal Engine 5 and train on rendered 2d frames)

[0] https://github.com/LiheYoung/Depth-Anything

[1] https://medium.com/@patriciogv/the-state-of-the-art-of-depth...

Very interesting work! More details here: https://depth-anything.github.io/

It seems better overall and per parameter than current work, with relative and absolute measurement.

Is there any research people are aware of that provides sub-mm level models? For 3D modeling purposes? Or is "classic" photogrammetry still the best option there?