MiDaS 3.1

• New models based on 5 different types of transformers (BEiT, Swin2, Swin, Next-ViT, LeViT)
• Training datasets extended from 10 to 12, including also KITTI and NYU Depth V2 using BTS split
• Best model, BEiT_{Large 512}, with resolution 512x512, is on average about 28% more accurate than MiDaS v3.0
• Integrated live depth estimation from camera feed

MiDaS v3 (DPT)

• New models based on Dense Prediction Transformers are on average 21% more accurate than MiDaS v2.1
• Additional DPT models can be found here

MiDaS v2.1

• New model that was trained on 10 datasets and is on average about 10% more accurate
• New light-weight model that achieves real-time performance on mobile platforms
• Sample applications for iOS and Android
• ROS package for easy deployment on robots

MiDaS v2

Missing dependency

MiDaS v1

Merge pull request #3 from barnjamin/master

adding configurable depth bits