Abstract <p>This paper is devoted to the use of Gabor filters to improve the efficiency of convolutional neural networks (CNNs) in image analysis, namely, image segmentation. The use of Gabor filters as an adaptive component in initial CNN layers, which improves the extraction of texture and structural features, is considered. To achieve the optimal tradeoff between the number of trainable parameters and accuracy, adaptive Gabor filters are proposed, which increase the number of input channels without significantly complicating the model. The comparative analysis of several architectures using PSPNet for image segmentation modified with adaptive Gabor filters is carried out. Constraints on the size of the filters that ensure acceptable computational costs are considered. The relevance of the proposed approach on an image segmentation dataset is confirmed, demonstrating an improvement in accuracy with the minimum increase in the number of parameters.</p>