Benchmarking Probabilistic Deep Learning Methods for License Plate Recognition
This addresses the issue of false predictions in automated license plate recognition for forensic and extreme environmental applications, though it is incremental as it benchmarks existing methods.
The paper tackles the problem of unreliable license plate recognition under out-of-distribution conditions by modeling prediction uncertainty, showing that uncertainty measures reliably detect false predictions and that a multi-task approach improves recognition performance by 109% and wrong prediction detection by 29%.
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 %.