Taking Acceleration to the Edge
Optical Flow is a technique to estimate the apparent motion of image objects caused by their movements in a video sequence.
Field Programmable Gate Arrays (FPGAs) are ideally suited for optical flow. Data collected can be used for pattern recognition, object sorting, robotic arm control, and more.
FPGAs can also be used as a vision processing accelerator inside the Edge Computing platform, to harness the power of artificial intelligence deep learning for analysis of the Optical Flow data.
AimValley Optical Flow Demonstrator
AimValley’s Optical Flow Demonstrator enables comparison of implementations running in software on a CPU and offloading with a graphics card and FPGA accelerator card. The design reads from a video file or live images from a webcam and writes outputs directly to a display or to a file.
The AimValley Optical Flow Demonstrator combines the Lucas Kanade Algorithm with Harris Corner Detection and color wheel encoding
System Configuration
- Standard x86 server PC
- FPGA card
- GPU card
- Open CV, OpenCL, C/C++ and HLS library components
Measurements Results
Ask for a demo
Applications
- Motion Detection
- Process automation
- Vision guided robotics
Benefits
- Acceleration - Performance Optimization
- Parallel Processing - Energy Efficiency
- Flexibility