indranildchandra/Deep-Learning-on-the-Edge

Deep-Learning-on-the-Edge

DL-on-the-Edge is a repository for keeping all the object detection demos for my talk at Google Developers Group (GDG) Mumbai and Bangalore 2019. Accessible and affordable compute power, volume and variety of data, better algorithms and frameworks are the 3 major reasons why Deep Learning as a field has exploded and has become so popular in the last decade. This talk was primarily focused on the first aspect - compute power and how modern DL based solutions can be deployed on edge devices with minimal compute power. Solutions like NUC (Next Unit of Computing) and NCS2 (Neural Compute Stick - 2) were discussed and how they can be leveraged to build solutions like auto-navigating robots and assistive devices with accompanying demos.

How-to Links:

1. Setup OpenVINO Toolkit - https://docs.openvinotoolkit.org/2019_R2/index.html
2. Setup NCS2 - https://software.intel.com/en-us/articles/get-started-with-neural-compute-stick
3. Setup Intel Realsense D435i Depth Camera - https://www.intelrealsense.com/videos-and-tutorials/

Links to other repositories containing other demos used in the talk:

• https://github.com/indranildchandra/Intel-Realsense-SLAM-Robotics
• https://github.com/laukik-hase/rsslam_ws

Other Demo Videos:

• Intel Realsense D435i Depth Camera - Basic Mapping Demo
• Intel Realsense D435i Depth Camera - IMU Pitch & Roll Demo
• Intel Realsense D435i Depth Camera - Mapping & Localization Demo
• Autonomous Quadruped Robot Demo