Papers › WaveMix: A Resource-efficient Neural Network for Image Analysis

WaveMix: A Resource-efficient Neural Network for Image Analysis

28 May 2022arXiv:2205.14375archive 2025-07-28

Pranav Jeevan, Kavitha Viswanathan, Anandu A S, Amit Sethi

We propose a novel neural architecture for computer vision -- WaveMix -- that is resource-efficient and yet generalizable and scalable. While using fewer trainable parameters, GPU RAM, and computations, WaveMix networks achieve comparable or better accuracy than the state-of-the-art convolutional neural networks, vision transformers, and token mixers for several tasks. This efficiency can translate to savings in time, cost, and energy. To achieve these gains we used multi-level two-dimensional discrete wavelet transform (2D-DWT) in WaveMix blocks, which has the following advantages: (1) It reorganizes spatial information based on three strong image priors -- scale-invariance, shift-invariance, and sparseness of edges -- (2) in a lossless manner without adding parameters, (3) while also reducing the spatial sizes of feature maps, which reduces the memory and time required for forward and backward passes, and (4) expanding the receptive field faster than convolutions do. The whole architecture is a stack of self-similar and resolution-preserving WaveMix blocks, which allows architectural flexibility for various tasks and levels of resource availability. WaveMix establishes new benchmarks for segmentation on Cityscapes; and for classification on Galaxy 10 DECals, Places-365, five EMNIST datasets, and iNAT-mini and performs competitively on other benchmarks. Our code and trained models are publicly available.

PaperPDFCode

Code

pranavphoenix/WaveMix officialmentioned in papermentioned on GitHubpytorchMIT report

Repository list and official/mentioned flags are the archive's, frozen 2025-07-28. Reachability, where shown, is from one Syntology probe window (2026-09-16 to 2026-09-18); repositories not probed show nothing. GitHub stars are not tracked.

Code Syntology ran Syntology

Not run by Syntology. Nothing on this page verifies that the listed code works.

Tasks

Efficient Neural NetworkImage ClassificationScene ClassificationSemantic SegmentationSpatial Token Mixer

1 archive task tag without a task page not shown.

Results from the paper archive 2025-07-28

TaskDatasetModelMetricValueRank at snapshotLeaderboardReport
Image Classification CIFAR-10 WaveMixLite-144/7 Percentage correct 97.29 #89 of 265 Archive leaderboard report
Image Classification CIFAR-100 WaveMixLite-256/7 Percentage correct 85.09 #68 of 211 Archive leaderboard report
Image Classification CIFAR-100 WaveMix-Lite-256/7 Percentage correct 70.20 #175 of 211 Archive leaderboard report
Image Classification Caltech-256 WaveMixLite-256/7 Accuracy 54.62 #5 of 5 Archive leaderboard report
Image Classification EMNIST-Balanced WaveMixLite-128/7 Accuracy 91.06 #3 of 20 Archive leaderboard report
Image Classification EMNIST-Byclass WaveMixLite-128/7 Accuracy 88.43 #1 of 1 Archive leaderboard report
Image Classification EMNIST-Bymerge WaveMixLite-128/16 Accuracy 91.80 #1 of 1 Archive leaderboard report
Image Classification EMNIST-Digits WaveMixLite-112/16 Accuracy (%) 99.82 #1 of 7 Archive leaderboard report
Image Classification EMNIST-Letters WaveMixLite-112/16 Accuracy 95.96 #1 of 11 Archive leaderboard report
Image Classification Fashion-MNIST WaveMixLite Percentage error 5.68 #6 of 34 Archive leaderboard report
Image Classification Galaxy10 DECals WaveMix PARAMS (M) 28 #1 of 2 Archive leaderboard report
Image Classification Galaxy10 DECals WaveMix Top-1 Accuracy (%) 95.42 #1 of 2 Archive leaderboard report
Image Classification ImageNet WaveMix-192/16 (level 3) Top 1 Accuracy 74.93% #964 of 1060 Archive leaderboard report
Image Classification Places365-Standard WaveMix-240/12 (level 4) Top 1 Accuracy 56.45 #4 of 4 Archive leaderboard report
Image Classification STL-10 WaveMixLite-256/7 Percentage correct 70.88 #89 of 117 Archive leaderboard report
Image Classification SVHN WaveMixLite-144/15 Percentage error 1.27 #5 of 62 Archive leaderboard report
Image Classification Tiny ImageNet Classification WaveMixLite-144/7 Validation Acc 77.47% #10 of 23 Archive leaderboard report
Image Classification iNat2021-mini WaveMix-256/16 (level 2) Top 1 Accuracy 61.75 #1 of 1 Archive leaderboard report
Image Classification mnist WaveMixLite Percentage error 0.25 #1 of 1 Archive leaderboard report
Scene Classification Places365-Standard WaveMix Top 1 Error 43.55 #1 of 2 Archive leaderboard report
Semantic Segmentation Cityscapes val WaveMix mIoU 82.7 #33 of 99 Archive leaderboard report
Semantic Segmentation Cityscapes val WaveMix-256/16 (Level-4) mIoU 82.60 #35 of 99 Archive leaderboard report

Ranks are positions in the archive's leaderboards as they stood at the 2025-07-28 snapshot. Results published since then are not among these rows, so a rank here is not a current standing.

Methods

2D DWTAverage PoolingDense ConnectionsDropoutGlobal Average PoolingLayer NormalizationMLP-MixerPoolFormerResidual Connection

Report a problem or propose a change · a person checks every report against the paper or source before anything changes; decisions are listed on /corrections