mlsanityData to 7 Oct 2026

MobileNetV4 -- Universal Models for the Mobile Ecosystem

arXiv 2404.10518 ↗First posted 16 Apr 2024cs.CVcited by 98

Not on the map. Why?

Abstract (arXiv)

We present the latest generation of MobileNets, known as MobileNetV4 (MNv4), featuring universally efficient architecture designs for mobile devices. At its core, we introduce the Universal Inverted Bottleneck (UIB) search block, a unified and flexible structure that merges Inverted Bottleneck (IB), ConvNext, Feed Forward Network (FFN), and a novel Extra Depthwise (ExtraDW) variant. Alongside UIB, we present Mobile MQA, an attention block tailored for mobile accelerators, delivering a significant 39% speedup. An optimized neural architecture search (NAS) recipe is also introduced which improves MNv4 search effectiveness. The integration of UIB, Mobile MQA and the refined NAS recipe results in a new suite of MNv4 models that are mostly Pareto optimal across mobile CPUs, DSPs, GPUs, as well as specialized accelerators like Apple Neural Engine and Google Pixel EdgeTPU - a characteristic not found in any other models tested. Finally, to further boost accuracy, we introduce a novel distillation technique. Enhanced by this technique, our MNv4-Hybrid-Large model delivers 87% ImageNet-1K accuracy, with a Pixel 8 EdgeTPU runtime of just 3.8ms.

Key facts. Cited by 98 papers in the map (data through 7 Oct 2026). Builds on 47 of its 57 references that are in the map. First posted on arXiv 16 Apr 2024.

Builds on

47 of 57 references.

References · 47

  1. 2017Attention Is All You Needcited by 59,932
  2. 2020An Image is Worth 16x16 Words: Transformers for Image Recognition at Scalecited by 29,915
  3. 2014Microsoft COCO: Common Objects in Contextcited by 17,498
  4. 2015Batch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shiftcited by 6,979
  5. 2019EfficientNet: Rethinking Model Scaling for Convolutional Neural Networkscited by 6,415
  6. 2016Layer Normalizationcited by 5,579
  7. 2015Rethinking the Inception Architecture for Computer Visioncited by 5,451
  8. 2022PaLM: Scaling Language Modeling with Pathwayscited by 5,304
  9. 2017mixup: Beyond Empirical Risk Minimizationcited by 5,155
  10. 2016Feature Pyramid Networks for Object Detectioncited by 5,021
  11. 2018MobileNetV2: Inverted Residuals and Linear Bottleneckscited by 4,773
  12. 2017Squeeze-and-Excitation Networkscited by 4,763
35 more
  1. 2017MobileNets: Efficient Convolutional Neural Networks for Mobile Vision Applicationscited by 3,827
  2. 2022A ConvNet for the 2020scited by 3,545
  3. 2016Gaussian Error Linear Units (GELUs)cited by 3,431
  4. 2019CutMix: Regularization Strategy to Train Strong Classifiers with Localizable Featurescited by 2,875
  5. 2019RandAugment: Practical automated data augmentation with a reduced search spacecited by 2,261
  6. 2019Searching for MobileNetV3cited by 1,984
  7. 2021Pyramid Vision Transformer: A Versatile Backbone for Dense Prediction without Convolutionscited by 1,960
  8. 2019Self-training with Noisy Student improves ImageNet classificationcited by 1,304
  9. 2017ShuffleNet: An Extremely Efficient Convolutional Neural Network for Mobile Devicescited by 1,194
  10. 2016Deep Networks with Stochastic Depthcited by 973
  11. 2018MnasNet: Platform-Aware Neural Architecture Search for Mobilecited by 937
  12. 2017Revisiting Unreasonable Effectiveness of Data in Deep Learning Eracited by 916
  13. 2019Once-for-All: Train One Network and Specialize it for Efficient Deploymentcited by 778
  14. 2019Fast Transformer Decoding: One Write-Head is All You Needcited by 700
  15. 2021MobileViT: Light-weight, General-purpose, and Mobile-friendly Vision Transformercited by 636
  16. 2018FBNet: Hardware-Aware Efficient ConvNet Design via Differentiable Neural Architecture Searchcited by 480
  17. 2019GhostNet: More Features from Cheap Operationscited by 410
  18. 2021CMT: Convolutional Neural Networks Meet Vision Transformerscited by 310
  19. 2021Knowledge distillation: A good teacher is patient and consistentcited by 255
  20. 2022EfficientFormer: Vision Transformers at MobileNet Speedcited by 240
  21. 2021Mobile-Former: Bridging MobileNet and Transformercited by 234
  22. 2023EfficientViT: Memory Efficient Vision Transformer with Cascaded Group Attentioncited by 172
  23. 2022Separable Self-attention for Mobile Vision Transformerscited by 163
  24. 2023FastViT: A Fast Hybrid Vision Transformer using Structural Reparameterizationcited by 161
  25. 2021Revisiting ResNets: Improved Training and Scaling Strategiescited by 158
  26. 2023Platypus: Quick, Cheap, and Powerful Refinement of LLMscited by 150
  27. 2018NetAdapt: Platform-Aware Neural Network Adaptation for Mobile Applicationscited by 146
  28. 2022Rethinking Vision Transformers for MobileNet Size and Speedcited by 121
  29. 2022MobileOne: An Improved One millisecond Mobile Backbonecited by 109
  30. 2020Can weight sharing outperform random architecture search? An investigation with TuNAScited by 68
  31. 2022Next-ViT: Next Generation Vision Transformer for Efficient Deployment in Realistic Industrial Scenarioscited by 65
  32. 2022GhostNetV2: Enhance Cheap Operation with Long-Range Attentioncited by 44
  33. 2020Discovering Multi-Hardware Mobile Models via Architecture Searchcited by 12
  34. 2022Searching for Efficient Neural Architectures for On-Device ML on Edge TPUscited by 4
  35. 2024On the Efficiency of Convolutional Neural Networkscited by 2

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2024: 12; 2025: 46; 2026: 40.

98 papers.

Sort:

By area:

  1. 2024reBEN: Refined BigEarthNet Dataset for Remote Sensing Image Analysiscited by 48
  2. 2024SnapGen: Taming High-Resolution Text-to-Image Models for Mobile Devices with Efficient Architectures and Trainingcited by 27
  3. 2024Active Data Curation Effectively Distills Large-Scale Multimodal Modelscited by 19
  4. 2024CAS-ViT: Convolutional Additive Self-attention Vision Transformers for Efficient Mobile Applicationscited by 12
  5. 2025iFormer: Integrating ConvNet and Transformer for Mobile Applicationcited by 9
  6. 2024NanoMVG: USV-Centric Low-Power Multi-Task Visual Grounding based on Prompt-Guided Camera and 4D mmWave Radarcited by 7
  7. 2025Advancing Earth Observation: A Survey on AI-Powered Image Processing in Satellitescited by 6
  8. 2025VL-RouterBench: A Benchmark for Vision-Language Model Routingcited by 6
  9. 2024Residual Kolmogorov-Arnold Network for Enhanced Deep Learningcited by 5
  10. 2024Semantic Edge Computing and Semantic Communications in 6G Networks: A Unifying Survey and Research Challengescited by 5
  11. 2024EMOv2: Pushing 5M Vision Model Frontiercited by 5
  12. 2025Fast and Accurate Gigapixel Pathological Image Classification with Hierarchical Distillation Multi-Instance Learningcited by 5
38 more
  1. 2025UltraFast-LiNET: Light-weight multi-scale shift convolutional network for real-time low-light image enhancementcited by 5
  2. 2025Thicker and Quicker: A Jumbo Token for Fast Plain Vision Transformerscited by 4
  3. 2025WeedSense: Multi-Task Learning for Weed Segmentation, Height Estimation, and Growth Stage Classificationcited by 4
  4. 2025VQualA 2025 Challenge on Face Image Quality Assessment: Methods and Resultscited by 4
  5. 2026PhyG-MoE: A Physics-Guided Mixture-of-Experts Framework for Energy-Efficient GNSS Interference Recognitioncited by 4
  6. 2026SF-Mamba: Rethinking State Space Model for Visioncited by 4
  7. 2025PatchRefiner V2: Fast and Lightweight Real-Domain High-Resolution Metric Depth Estimationcited by 3
  8. 2025Cycle Training with Semi-Supervised Domain Adaptation: Bridging Accuracy and Efficiency for Real-Time Mobile Scene Detectioncited by 3
  9. 2025MoNetV2: Enhanced Motion Network for Freehand 3D Ultrasound Reconstructioncited by 3
  10. 2025MoCom: Motion-based Inter-MAV Visual Communication Using Event Vision and Spiking Neural Networkscited by 3
  11. 2025ZIP: Scalable Crowd Counting via Zero-Inflated Poisson Modelingcited by 2
  12. 2025DeDelayed: Deleting Remote Inference Delay via On-Device Correctioncited by 2
  13. 2025IDEAL-M3D: Instance Diversity-Enriched Active Learning for Monocular 3D Detectioncited by 2
  14. 2025HyperVL: An Efficient and Dynamic Multimodal Large Language Model for Edge Devicescited by 2
  15. 2026Distilling Lightweight Domain Experts from Large ML Models by Identifying Relevant Subspacescited by 2
  16. 2026EgoPoseFormer v2: Accurate Egocentric Human Motion Estimation for AR/VRcited by 2
  17. 2026LiteFrame: Efficient Vision Encoders Unlock Frame Scaling in Video LLMscited by 2
  18. 2024AFIDAF: Alternating Fourier and Image Domain Adaptive Filters as an Efficient Alternative to Attention in ViTscited by 1
  19. 2025HGO-YOLO: Advancing Anomaly Behavior Detection with Hierarchical Features and Lightweight Optimized Detectioncited by 1
  20. 2025An improved EfficientNetV2 for garbage classificationcited by 1
  21. 2025DMAT: An End-to-End Framework for Joint Atmospheric Turbulence Mitigation and Object Detectioncited by 1
  22. 2025RulerNet: Learning Perspective-Invariant Ruler Representations for Robust Image Scale Estimationcited by 1
  23. 2025Mobile-Friendly Deep Learning for Plant Disease Detection: A Lightweight CNN Benchmark Across 101 Classes of 33 Cropscited by 1
  24. 2025TSLA: A Task-Specific Learning Adaptation for Semantic Segmentation on Autonomous Vehicles Platformcited by 1
  25. 2025Efficient Test-Time Adaptation through Latent Subspace Coefficients Searchcited by 1
  26. 2025MOBIUS: Big-to-Mobile Universal Instance Segmentation via Multi-modal Bottleneck Fusion and Calibrated Decoder Pruningcited by 1
  27. 2026Investigating the Robustness of Subtask Distillation under Spurious Correlationcited by 1
  28. 2026Surveillance Facial Image Quality Assessment: A Multi-dimensional Dataset and Lightweight Modelcited by 1
  29. 2026PicoSAM3: Real-Time In-Sensor Region-of-Interest Segmentationcited by 1
  30. 2026CarbonEdge: Carbon-Aware Deep Learning Inference Framework for Sustainable Edge Computingcited by 1
  31. 2026Adding Thermal Awareness to Visual Systems in Real-Time via Distilled Diffusion Modelscited by 1
  32. 2024Vision-Based Power Line Cables and Pylons Detection for Low Flying Aircraftcited by 0
  33. 2024OCDet: Object Center Detection via Bounding Box-Aware Heatmap Prediction on Edge Devices with NPUscited by 0
  34. 2024CLIP-PING: Boosting Lightweight Vision-Language Models with Proximus Intrinsic Neighbors Guidancecited by 0
  35. 2025Robust Real-Time Endoscopic Stereo Matching under Fuzzy Tissue Boundariescited by 0
  36. 2025SSVQ: Unleashing the Potential of Vector Quantization with Sign-Splittingcited by 0
  37. 2025Lightweight Models for Emotional Analysis in Videocited by 0
  38. 2025GmNet: Revisiting Gating Mechanisms From A Frequency Viewcited by 0

50 most cited of 98 All 98 (JSON)