YOLOv3 Matlab Implementation for Robocup challenge.
- Design a YOLOv2 Architecture for Robocup 2021 Manipulator challenge, that recognizes bottles and cans in a simulated environment.
- Train the neural network for a robust prediction.
It's required to have Image Processing Toolbox
, Computer Vision Toolbox
and Deep Learning Toolbox
Multiple images were taken in the Gazebo environment from the perspective. of the RGB-D Camera of the Genova3 Robot Arm. The input image is a 480x270x3 (RGB) resized to 270x270x3. A DataStore is created for the Labels and the Images , with the Image Labeler Matlab App
Based on a SqueezeNet Feature extraction.
## Authors CEIMTUN-RAS 2021 - Universidad Nacional de Colombia.