Implement HybridNets: End-to-End Perception Network

[chainyo]

🌟 New model addition

Model description

Quote from the Github repository:

HybridNets is an end2end perception network for multi-tasks.    
Our work focused on traffic object detection, drivable area segmentation and lane detection.   
HybridNets can run real-time on embedded systems, and obtains SOTA Object Detection, 
Lane Detection on BDD100K Dataset.

Open source status

• [x] the model implementation is available: yes, they have pushed everything here on Github
• [x] the model weights are available: yes, there are available directly here on Github.
• [x] who are the authors: Dat Vu and Bao Ngo and Hung Phan (@xoiga123 @datvuthanh)

Btw, I would love to help to implement this kind of models to the Hub if possible :hugs: . But I don't know if it fits to Transformers library or not and if it involves a lot of work to implement it, but I want to try to help!

[patil-suraj]

cc @NielsRogge

[amyeroberts]

Hi @chainyo, thanks for opening this new model request!

Here's a guide on adding a model to the hub if you, or anyone else in the community is still interested: https://huggingface.co/docs/transformers/custom_models