Increasing the Accuracy of a Neural Network Using Frequency Selective Mesh-to-Grid Resampling
This addresses the problem of neural network sensitivity to minor input changes for researchers and practitioners in computer vision, though it is incremental as it builds on existing interpolation methods.
The paper tackles the understudied influence of image interpolation on neural network performance by proposing frequency selective mesh-to-grid resampling (FSMR) for input data processing, showing that it can increase classification accuracy by up to 4.31 percentage points for ResNet50 on the Oxflower17 dataset.
Neural networks are widely used for almost any task of recognizing image content. Even though much effort has been put into investigating efficient network architectures, optimizers, and training strategies, the influence of image interpolation on the performance of neural networks is not well studied. Furthermore, research has shown that neural networks are often sensitive to minor changes in the input image leading to drastic drops of their performance. Therefore, we propose the use of keypoint agnostic frequency selective mesh-to-grid resampling (FSMR) for the processing of input data for neural networks in this paper. This model-based interpolation method already showed that it is capable of outperforming common interpolation methods in terms of PSNR. Using an extensive experimental evaluation we show that depending on the network architecture and classification task the application of FSMR during training aids the learning process. Furthermore, we show that the usage of FSMR in the application phase is beneficial. The classification accuracy can be increased by up to 4.31 percentage points for ResNet50 and the Oxflower17 dataset.