Goswami-Pratik / Airborne-Object-Sense-Avoid

Airborne object sense & avoid, project that uses a machine learning model to infer images to detect airborne objects, which would then be classified and tracked to predict a manoeuvre that avoids the obstacles
MIT License
2 stars 0 forks source link

which tracking method did you use #7

Open jeannotes opened 2 years ago

Goswami-Pratik commented 2 years ago

Uses a simple tracking approach where the object's relative position on the frame is monitored and predicts an opposite direction. Inspired by the Tracking Objects as Points paper. In future, working to integrate a form of SiamMOT (Siamese Multi-Object Tracking).