{"url":"/method/res2net-block","slug":"res2net-block","name":"Res2Net Block","full_name":"Res2Net Block","full_name_withheld":false,"description_markdown":"A **Res2Net Block** is an image model block that constructs hierarchical residual-like connections\r\nwithin one single [residual block](https://paperswithcode.com/method/residual-block). It was proposed as part of the [Res2Net](https://paperswithcode.com/method/res2net) CNN architecture.\r\n\r\nThe block represents multi-scale features at a granular level and increases the range of receptive fields for each network layer. The $3 \\times 3$ filters of $n$ channels is replaced with a set of smaller filter groups, each with $w$ channels. These smaller filter groups are connected in a hierarchical residual-like style to increase the number of scales that the output features can represent. Specifically, we divide input feature maps into several groups. A group of filters first extracts features from a group of input feature maps. Output features of the previous group are then sent to the next group of filters along with another group of input feature maps. \r\n\r\nThis process repeats several times until all input feature maps are processed. Finally, feature maps from all groups are concatenated and sent to another group of $1 \\times 1$ filters to fuse information altogether. Along with any possible path in which input features are transformed to output features, the equivalent receptive field increases whenever it passes a $3 \\times 3$ filter, resulting in many equivalent feature scales due to combination effects.\r\n\r\nOne way of thinking of these blocks is that they expose a new dimension, **scale**,  alongside the existing dimensions of depth, width, and cardinality.","description_state":"present","introduced_year":null,"introduced_by":{"title":"Res2Net: A New Multi-scale Backbone Architecture","paper":"/paper/res2net-a-new-multi-scale-backbone","first_author":"Shang-Hua Gao","n_authors":6,"url_abs":null,"archive_paper_url":"https://paperswithcode.com/paper/res2net-a-new-multi-scale-backbone"},"source":{"url":"https://arxiv.org/abs/1904.01169v3","title":"Res2Net: A New Multi-scale Backbone Architecture","url_on_a_paper_host":true},"code_snippet_url":"https://github.com/Res2Net/Res2Net-PretrainedModels/blob/d56af39dfc977bcbf3216871d20e2a9e16801c8a/res2net.py#L20","code_snippet_url_on_a_code_host":true,"categories":[{"area":"Computer Vision","area_id":"computer-vision","collection":"Image Model Blocks","url":"/methods/category/image-model-blocks","pwc_aliases":[]},{"area":"General","area_id":"general","collection":"Skip Connection Blocks","url":"/methods/category/skip-connection-blocks","pwc_aliases":[]}],"n_papers_tagged":25,"archive_num_papers":25,"papers_newest_first":[{"paper":"/paper/nes2net-a-lightweight-nested-architecture-for","title":"Nes2Net: A Lightweight Nested Architecture for Foundation Model Driven Speech Anti-spoofing","date":"2025-04-08","arxiv_id":"2504.05657","n_code_links":1,"syntology":null},{"paper":null,"title":"ERes2NetV2: Boosting Short-Duration Speaker Verification Performance with Computational Efficiency","date":"2024-06-04","arxiv_id":"2406.02167","n_code_links":0,"syntology":null},{"paper":null,"title":"Speech enhancement deep-learning architecture for efficient edge processing","date":"2024-05-27","arxiv_id":"2405.16834","n_code_links":0,"syntology":null},{"paper":"/paper/enhancing-retinal-vascular-structure","title":"Enhancing Retinal Vascular Structure Segmentation in Images With a Novel Design Two-Path Interactive Fusion Module Model","date":"2024-03-03","arxiv_id":"2403.01362","n_code_links":1,"syntology":null},{"paper":"/paper/next-tdnn-modernizing-multi-scale-temporal","title":"NeXt-TDNN: Modernizing Multi-Scale Temporal Convolution Backbone for Speaker Verification","date":"2023-12-14","arxiv_id":"2312.08603","n_code_links":1,"syntology":null},{"paper":null,"title":"Improving Short Utterance Anti-Spoofing with AASIST2","date":"2023-09-15","arxiv_id":"2309.08279","n_code_links":0,"syntology":null},{"paper":null,"title":"A region and category confidence-based multi-task network for carotid ultrasound image segmentation and classification","date":"2023-07-02","arxiv_id":"2307.00583","n_code_links":0,"syntology":null},{"paper":null,"title":"Multi-perspective Information Fusion Res2Net with RandomSpecmix for Fake Speech Detection","date":"2023-06-27","arxiv_id":"2306.15389","n_code_links":0,"syntology":null},{"paper":"/paper/an-enhanced-res2net-with-local-and-global","title":"An Enhanced Res2Net with Local and Global Feature Fusion for Speaker Verification","date":"2023-05-22","arxiv_id":"2305.12838","n_code_links":2,"syntology":null},{"paper":"/paper/breaking-through-the-haze-an-advanced-non","title":"Breaking Through the Haze: An Advanced Non-Homogeneous Dehazing Method based on Fast Fourier Convolution and ConvNeXt","date":"2023-05-08","arxiv_id":"2305.04430","n_code_links":1,"syntology":null},{"paper":"/paper/cross-modal-information-fusion-for-voice","title":"Cross-modal information fusion for voice spoofing detection","date":"2023-02-01","arxiv_id":null,"n_code_links":1,"syntology":null},{"paper":null,"title":"Synthetic Voice Detection and Audio Splicing Detection using SE-Res2Net-Conformer Architecture","date":"2022-10-07","arxiv_id":"2210.03581","n_code_links":0,"syntology":null},{"paper":null,"title":"Breast Cancer Classification Based on Histopathological Images Using a Deep Learning Capsule Network","date":"2022-08-01","arxiv_id":"2208.00594","n_code_links":0,"syntology":null},{"paper":"/paper/selective-kernel-attention-for-robust-speaker","title":"Frequency and Multi-Scale Selective Kernel Attention for Speaker Verification","date":"2022-04-03","arxiv_id":"2204.01005","n_code_links":1,"syntology":null},{"paper":"/paper/mfsnet-a-multi-focus-segmentation-network-for","title":"MFSNet: A Multi Focus Segmentation Network for Skin Lesion Segmentation","date":"2022-03-27","arxiv_id":"2203.14341","n_code_links":2,"syntology":null},{"paper":null,"title":"Phase-Aware Spoof Speech Detection Based on Res2Net with Phase Network","date":"2022-03-21","arxiv_id":"2203.10793","n_code_links":0,"syntology":null},{"paper":"/paper/raw-waveform-speaker-verification-for","title":"Pushing the limits of raw waveform speaker recognition","date":"2022-03-16","arxiv_id":"2203.08488","n_code_links":2,"syntology":null},{"paper":"/paper/channel-wise-gated-res2net-towards-robust","title":"Channel-wise Gated Res2Net: Towards Robust Detection of Synthetic Speech Attacks","date":"2021-07-19","arxiv_id":"2107.08803","n_code_links":2,"syntology":null},{"paper":"/paper/dw-gan-a-discrete-wavelet-transform-gan-for","title":"DW-GAN: A Discrete Wavelet Transform GAN for NonHomogeneous Dehazing","date":"2021-04-18","arxiv_id":"2104.08911","n_code_links":1,"syntology":null},{"paper":"/paper/partially-connected-differentiable","title":"Partially-Connected Differentiable Architecture Search for Deepfake and Spoofing Detection","date":"2021-04-07","arxiv_id":"2104.03123","n_code_links":1,"syntology":null},{"paper":null,"title":"The xx205 System for the VoxCeleb Speaker Recognition Challenge 2020","date":"2020-10-31","arxiv_id":"2011.00200","n_code_links":0,"syntology":null},{"paper":"/paper/replay-and-synthetic-speech-detection-with","title":"Replay and Synthetic Speech Detection with Res2net Architecture","date":"2020-10-28","arxiv_id":"2010.15006","n_code_links":2,"syntology":null},{"paper":"/paper/gated-res2net-for-multivariate-time-series","title":"Gated Res2Net for Multivariate Time Series Analysis","date":"2020-09-19","arxiv_id":"2009.11705","n_code_links":1,"syntology":null},{"paper":null,"title":"Improved Res2Net model for Person re-identification","date":"2019-10-08","arxiv_id":"1910.04061","n_code_links":0,"syntology":null},{"paper":"/paper/res2net-a-new-multi-scale-backbone","title":"Res2Net: A New Multi-scale Backbone Architecture","date":"2019-04-02","arxiv_id":"1904.01169","n_code_links":34,"syntology":{"ran":3,"of":9,"unverified":6,"pointer_only":9}}],"papers_shown":25,"tasks":[{"task":"/task/speaker-verification","name":"Speaker Verification","papers":9},{"task":"/task/segmentation","name":"Segmentation","papers":3},{"task":"/task/face-swapping","name":"Face Swapping","papers":2},{"task":"/task/semantic-segmentation","name":"Semantic Segmentation","papers":2},{"task":"/task/speaker-recognition","name":"Speaker Recognition","papers":2},{"task":"/task/automatic-speech-recognition-2","name":"Automatic Speech Recognition","papers":1},{"task":"/task/binary-classification","name":"Binary Classification","papers":1},{"task":"/task/cancer-classification","name":"Cancer Classification","papers":1},{"task":"/task/classification-1","name":"Classification","papers":1},{"task":"/task/computational-efficiency","name":"Computational Efficiency","papers":1},{"task":"/task/data-augmentation","name":"Data Augmentation","papers":1},{"task":"/task/decoder","name":"Decoder","papers":1},{"task":"/task/deep-learning","name":"Deep Learning","papers":1},{"task":"/task/deepfake-detection","name":"DeepFake Detection","papers":1},{"task":"/task/dimensionality-reduction","name":"Dimensionality Reduction","papers":1},{"task":"/task/feature-engineering","name":"Feature Engineering","papers":1},{"task":null,"name":"GPU","papers":1},{"task":null,"name":"Generative Adversarial Network","papers":1},{"task":"/task/graph-attention","name":"Graph Attention","papers":1},{"task":"/task/image-classification","name":"Image Classification","papers":1}],"tasks_shown":20,"n_tasks":45,"usage_by_year":[{"year":"2019","papers":2},{"year":"2020","papers":3},{"year":"2021","papers":3},{"year":"2022","papers":6},{"year":"2023","papers":7},{"year":"2024","papers":3},{"year":"2025","papers":1}],"row_source":"methods_table","archive":{"source":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","archive_url":"https://paperswithcode.com/method/res2net-block"},"syntology_read_at":"2026-09-24T18:15:14+00:00"}