HAT Super-Resolution and a PARSeq+CLIP4STR Voting Ensemble for Extreme In-the-Wild License Plate Recognition
2026-07-09 • Computer Vision and Pattern Recognition
Computer Vision and Pattern Recognition
AI summaryⓘ
The authors created a system to improve blurry or tiny license plate images so that the characters are easier to read. They first use a special type of AI called a Hybrid Attention Transformer to enhance the images. Then, two different text recognition models work together to read the characters, and if the system isn’t confident about a character, it chooses to skip it rather than guess. Their approach focuses on only trying to improve parts of the image that actually need it, and it runs fast enough to meet the challenge’s time limits.
Super-resolutionHybrid Attention TransformerScene-text recognitionConfidence weightingCharacter votingImage legibilityAsymmetric scoringEnsemble modelsRTX 3090 GPUXLPSR challenge
Authors
Karthik Sivarama Krishnan, Koushik Sivarama Krishnan
Abstract
We describe our entry to the ICIP 2026 Grand Challenge on Extreme In-the-Wild License Plate Super-Resolution (XLPSR), which scored 9.73 wECR on the public validation leaderboard. The system pairs a Hybrid Attention Transformer super-resolution (HAT) front-end with an ensemble of two scene-text recognisers (PARSeq-S and CLIP4STR-B) and a confidence-weighted character-voting scheme that abstains on uncertain positions. We treat XLPSR as a recognition task gated by image legibility: the SR step exists to lift characters out of sub-pixel territory, and the asymmetric scoring rule (+2 / -1 / 0) is exploited explicitly through abstention. Our pipeline runs in 1.7 s per sequence on RTX 3090 (max 2.7 s, p99 2.4 s), well under the 60 s/sequence Docker budget.