Summary
Video super-resolution methods usually get judged by pixel-level scores, but this does not always match how well you can recognize important details like faces and license plates. The authors created a new way to test super-resolution that focuses on how well these methods help cameras recognize faces and license plates in surveillance videos. They made a dataset with challenging low-resolution videos affected by realistic distortions and measure recognition accuracy after super-resolution. Their experiments show that some super-resolution techniques improve recognition rates, and they provide tools for others to test their methods using these meaningful recognition scores.
What this means in practice
- •For security camera operators: Assess video enhancement methods based on actual improvements in identifying faces and license plates in poor-quality surveillance footage.
- •For video analytics developers: Develop video super-resolution algorithms optimized for boosting recognition accuracy under real-world video degradations beyond simple resizing.
Authors
Kavitha Viswanathan, Vrinda Goel, Shlesh Gholap, Devayan Ghosh, Madhav Gupta, Dhruvi Ganatra, Sanket Potdar, Amit Sethi
Abstract
Video super-resolution (VSR) is normally judged by PSNR and SSIM on clips that were downsampled bicubically, although in surveillance its purpose is to make faces and licence plates \emph{recognisable}. We present FANVIDv2, a benchmark that scores VSR by what a recognition pipeline can do with its output. FANVIDv2 provides $320\times180$ low-resolution (LR) clips with high-resolution (HR) references for 48 public figures (with one HR gallery image each) and 375 licence-plate clips covering 360 distinct plate strings. LR clips are generated with a randomised compound degradation (blur, resize jitter, sensor noise, JPEG compression, final downsampling) rather than bicubic downsampling alone. Two metrics score recognition \emph{inside} detections: FaceRecBox rewards a face only if it is localised and correctly identified, and TextRecBox scores plate transcriptions by normalised edit distance weighted by localisation quality. With a 2.3\,M-parameter VSR baseline (RCDM), FaceRecBox rises from 0.6864 to 0.7222, identity accuracy on matched faces from 84.35\% to 86.93\%, and TextRecBox from 0.3088 to 0.3667; a residual-map gated variant (RCDM-RMGF) reaches 0.3801 on plates. We describe the degradation model, the baseline architectures and the scorers in detail, and release annotations, metadata, download and degradation scripts and evaluation code.