Benchmarking Probabilistic Deep Learning Methods for License Plate Recognition

TL;DR

This paper evaluates probabilistic deep learning methods for license plate recognition, emphasizing the importance of uncertainty modeling to detect errors, especially under challenging conditions, and demonstrates significant performance improvements.

Contribution

It introduces a comparison of uncertainty quantification methods in license plate recognition and shows how combining classification with super-resolution enhances accuracy and error detection.

Findings

Uncertainty measures reliably identify wrong predictions.

Multi-task learning improves recognition accuracy by 109%.

Detection of errors increases by 29% with combined methods.

Abstract

Learning-based algorithms for automated license plate recognition implicitly assume that the training and test data are well aligned. However, this may not be the case under extreme environmental conditions, or in forensic applications where the system cannot be trained for a specific acquisition device. Predictions on such out-of-distribution images have an increased chance of failing. But this failure case is oftentimes hard to recognize for a human operator or an automated system. Hence, in this work we propose to model the prediction uncertainty for license plate recognition explicitly. Such an uncertainty measure allows to detect false predictions, indicating an analyst when not to trust the result of the automated license plate recognition. In this paper, we compare three methods for uncertainty quantification on two architectures. The experiments on synthetic noisy or blurred low-resolution images show that the predictive uncertainty reliably finds wrong predictions. We also show that a multi-task combination of classification and super-resolution improves the recognition performance by 109\% and the detection of wrong predictions by 29 %.