Cross-Dataset-Robust Method for Blind Real-World Image Quality Assessment
This addresses the challenge of accurately assessing visual quality for arbitrary real-world images, which is crucial for applications in image processing and computer vision, though it is an incremental improvement over existing BIQA methods.
The paper tackles the problem of blind image quality assessment (BIQA) models overfitting to specific training datasets, proposing a cross-dataset-robust method that uses multiple Swin-Transformer models to generate pseudo-labels from a large-scale dataset of 1,000,000 image pairs, achieving better performance than some state-of-the-art methods in cross-dataset tests.
Although many effective models and real-world datasets have been presented for blind image quality assessment (BIQA), recent BIQA models usually tend to fit specific training set. Hence, it is still difficult to accurately and robustly measure the visual quality of an arbitrary real-world image. In this paper, a robust BIQA method, is designed based on three aspects, i.e., robust training strategy, large-scale real-world dataset, and powerful backbone. First, many individual models based on popular and state-of-the-art (SOTA) Swin-Transformer (SwinT) are trained on different real-world BIQA datasets respectively. Then, these biased SwinT-based models are jointly used to generate pseudo-labels, which adopts the probability of relative quality of two random images instead of fixed quality score. A large-scale real-world image dataset with 1,000,000 image pairs and pseudo-labels is then proposed for training the final cross-dataset-robust model. Experimental results on cross-dataset tests show that the performance of the proposed method is even better than some SOTA methods that are directly trained on these datasets, thus verifying the robustness and generalization of our method.