Lightweight Image Super-Resolution with ConvNeXt Residual Network

Single image super-resolution based on convolutional neural networks has been very successful in recent years. However, as the computational cost is too high, making it difficult to apply resource-constrained devices, a big challenge for existing approaches find balance between complexity of CNN model and quality resulting SR. To solve this problem, various lightweight SR have proposed. In paper, we propose efficient residual (IRN), which differ from previous that aggregate more powerful features by improving feature utilization through complex layer-connection strategies. The main idea simplify aggregation using simple modules learning, thus achieving good trade-off addition, revisit impact activation function observe different functions an performance model. experiment results show IRN outperforms state-of-the-art methods benchmark tests while maintaining relatively low cost. code will be available at https://github.com/kptx666/IRN .