nnMobileNet++: Towards Efficient Hybrid Networks for Retinal Image Analysis

TL;DR

nnMobileNet++ is a hybrid neural network architecture combining convolutional and transformer components, designed to improve retinal image analysis by capturing both local and global features efficiently.

Contribution

It introduces a novel hybrid architecture with dynamic snake convolution and stage-specific transformers, enhancing retinal analysis performance over previous lightweight models.

Findings

Achieves state-of-the-art accuracy on retinal datasets

Maintains low computational cost and efficiency

Demonstrates improved modeling of vascular features

Abstract

Retinal imaging is a critical, non-invasive modality for the early detection and monitoring of ocular and systemic diseases. Deep learning, particularly convolutional neural networks (CNNs), has significant progress in automated retinal analysis, supporting tasks such as fundus image classification, lesion detection, and vessel segmentation. As a representative lightweight network, nnMobileNet has demonstrated strong performance across multiple retinal benchmarks while remaining computationally efficient. However, purely convolutional architectures inherently struggle to capture long-range dependencies and model the irregular lesions and elongated vascular patterns that characterize on retinal images, despite the critical importance of vascular features for reliable clinical diagnosis. To further advance this line of work and extend the original vision of nnMobileNet, we propose nnMobileNet++, a hybrid architecture that progressively bridges convolutional and transformer representations. The framework integrates three key components: (i) dynamic snake convolution for boundary-aware feature extraction, (ii) stage-specific transformer blocks introduced after the second down-sampling stage for global context modeling, and (iii) retinal image pretraining to improve generalization. Experiments on multiple public retinal datasets for classification, together with ablation studies, demonstrate that nnMobileNet++ achieves state-of-the-art or highly competitive accuracy while maintaining low computational cost, underscoring its potential as a lightweight yet effective framework for retinal image analysis.