A near real-time detection system to track physical objects using FPGA technology and a camera for enhanced image processing.
This project was completed for EC 551 at Boston University.
Team Members:
- Mehedi Hassan
- Visaal Nekkanti
- Andrew Woska
- Abin George
- Getting Started
- Code To Look At
- Hardware
- Peripherals
- Licenses and Contributions
- Video
- Known Problems
The Verilog files for out implementation are located in Sources along with the ESP32-CAM implementation.
The Xilinx IP and Vivado files are located in Vivado Files. Since we are using the Basys 3, make sure to use the correct directory. The Nexys implementation was used for simulation purposes. The hardware layout is given in a later section.
For a general understanding of this program, feel free to look at the presentation and the demo video.
The image processing algorithm is located in Sources/Full Integration - Basys/houghAccumulator3.v or
here.
This project requires two pieces of hardware:
- FPGA
- Basys 3 was used for implementation
- Nexys A7 was used for testing but has a different constraint file.
- Camera
- ESP32-CAM was used for the initial implementation
- OV7670 is used for the current implementation
All fully integrated modules use a BRAM IP, which may not transfer well when pulling this repo. For this reason, the following is an image of the IP block.
This project uses parts of the Xilinx IP.
This project is licensed here.
The demo video is located here on Google Drive. You must have the correct access to get to it. Reach out to Andrew at agwoska@bu.edu or andrew@woska.org for access if needed.
Title: DetectoVision: Object Detection on FPGA
Description:
A near real-time detection system to track physical objects using FPGA technology and a camera for enhanced image processing.
This project was created for EC551 at Boston University in Fall 2023.
See our GitHub for more information at: https://github.com/agwoska/DetectoVision
Team Members:
- Mehedi Hassan
- Visaal Nekkanti
- Andrew Woska
- Abin George
- The ESP32-CAM does not always take an uncorrupted image
