btx0424 / OmniDrones

https://omnidrones.readthedocs.io/
MIT License
130 stars 24 forks source link

Vision-based training with MobileNetV3 as encoder #13

Closed UltronAI closed 1 year ago

UltronAI commented 1 year ago

Add vision-based training. Use pretrained MobileNetV3Small (or Large) for depth (copy channels three times). I didn't check if it's compatible with the latest commit. You can review the code and merge them into your branch if needed.