Project Details

[Return to Previous Page]

Edge-Optimized Image Upscaling

Company: SME Solutions Inc

Major(s):
Primary: CMPEN
Secondary: CMPSC

Non-Disclosure Agreement: NO

Intellectual Property: YES

The senior design team will develop a working prototype that captures low-resolution video from a camera, compresses it in real time using the high-performance libjpeg-turbo codec, and transmits it to a second embedded system over a constrained communication link. On the receiving end, the system will decode the JPEG stream and pass the images through a lightweight AI-based super-resolution module to upscale the video for display. This will demonstrate an effective reduction in transmission bandwidth while preserving important visual details for end-user consumption. The solution will be optimized for edge-class hardware, where processing resources are limited. The team will evaluate and implement a multi-processor offloading strategy, leveraging CPU, GPU, or hardware accelerators, to sustain real-time performance of at least 30 frames per second. The final deliverables will include bandwidth and quality benchmarks (e.g., bitrate savings, PSNR/SSIM scores, and latency measurements), as well as deployment scripts and documentation for replicating the pipeline on typical ARM-based platforms.