In this repository, the architectures are introduced for 3D object Recognition for RGB-D images for Robotic Grasping purpose .
The main goal of this project is to create an Deep Convolutional Neural Network Architecture for 3D Object Detection for RGB-D Images that is well documented for those who are interested in a understanding of the architectural design "https://digital.library.txstate.edu/handle/10877/13036". The provided documentation hopefully make it easier to understand the application, and code provided in the repository.
The repository currently provides the following network architectures:
Architecture I.py
Architecture II.py
Architecture III.py
Architecture IV.py
All these Multimodal architectures are trained from scratch on LEAP server(Linux Environment)
You can also use this architectures as a base network for SSD network for object detection bounding boxes prediction.