mlsanityData to 7 Oct 2026

Efficient Universal Perception Encoder

arXiv 2603.22387 ↗First posted 23 Mar 2026cs.CVcited by 6

Not on the map. Why?

Abstract (arXiv)

Running AI models on smart edge devices can unlock versatile user experiences, but presents challenges due to limited compute and the need to handle multiple tasks simultaneously. This requires a vision encoder with small size but powerful and versatile representations. We present our method, Efficient Universal Perception Encoder (EUPE), which offers both inference efficiency and universally good representations for diverse downstream tasks. We achieve this by distilling from multiple domain-expert foundation vision encoders. Unlike previous agglomerative methods that directly scale down from multiple teachers to an efficient encoder, we demonstrate the importance of first scaling up to a large proxy teacher and then scaling down from this single teacher. Experiments show that EUPE achieves on-par or better performance than individual domain experts of the same size on diverse task domains and also outperforms previous agglomerative encoders. We release the full family of EUPE models and the code to foster future research.

Key facts. Cited by 6 papers in the map (data through 7 Oct 2026). Builds on 44 of its 54 references that are in the map. First posted on arXiv 23 Mar 2026.

Builds on

44 of 54 references.

References · 44

  1. 2015Deep Residual Learning for Image Recognitioncited by 51,809
  2. 2021Learning Transferable Visual Models From Natural Language Supervisioncited by 33,456
  3. 2020An Image is Worth 16x16 Words: Transformers for Image Recognition at Scalecited by 29,915
  4. 2014Microsoft COCO: Common Objects in Contextcited by 17,498
  5. 2021Swin Transformer: Hierarchical Vision Transformer using Shifted Windowscited by 12,082
  6. 2015Distilling the Knowledge in a Neural Networkcited by 10,076
  7. 2023Segment Anythingcited by 8,158
  8. 2023Visual Instruction Tuningcited by 7,792
  9. 2023DINOv2: Learning Robust Visual Features without Supervisioncited by 7,764
  10. 2021Masked Autoencoders Are Scalable Vision Learnerscited by 7,339
  11. 2021Emerging Properties in Self-Supervised Vision Transformerscited by 6,893
  12. 2016Densely Connected Convolutional Networkscited by 6,691
32 more
  1. 2020Training data-efficient image transformers & distillation through attentioncited by 4,560
  2. 2022A ConvNet for the 2020scited by 3,545
  3. 2023Sigmoid Loss for Language Image Pre-Trainingcited by 3,048
  4. 2024SAM 2: Segment Anything in Images and Videoscited by 2,784
  5. 2016Aggregated Residual Transformations for Deep Neural Networkscited by 2,540
  6. 2021BEiT: BERT Pre-Training of Image Transformerscited by 2,360
  7. 2022Learn to Explain: Multimodal Reasoning via Thought Chains for Science Question Answeringcited by 1,881
  8. 2019Towards VQA Models That Can Readcited by 1,696
  9. 2023Evaluating Object Hallucination in Large Vision-Language Modelscited by 1,696
  10. 2023MME: A Comprehensive Evaluation Benchmark for Multimodal Large Language Modelscited by 1,624
  11. 2021An Empirical Study of Training Self-Supervised Vision Transformerscited by 1,610
  12. 2025DINOv3cited by 1,297
  13. 2025SigLIP 2: Multilingual Vision-Language Encoders with Improved Semantic Understanding, Localization, and Dense Featurescited by 1,127
  14. 2025SAM 3: Segment Anything with Conceptscited by 835
  15. 2024Cambrian-1: A Fully Open, Vision-Centric Exploration of Multimodal LLMscited by 822
  16. 2019Improved Knowledge Distillation via Teacher Assistantcited by 567
  17. 2019A Comprehensive Overhaul of Feature Distillationcited by 341
  18. 2025Perception Encoder: The best visual embeddings are not at the output of the networkcited by 315
  19. 2023Demystifying CLIP Datacited by 265
  20. 2023EfficientSAM: Leveraged Masked Image Pretraining for Efficient Segment Anythingcited by 187
  21. 2023AM-RADIO: Agglomerative Vision Foundation Model -- Reduce All Domains Into Onecited by 165
  22. 2019SPair-71k: A Large-scale Benchmark for Semantic Correspondencecited by 140
  23. 2024Multimodal Autoregressive Pre-training of Large Vision Encoderscited by 108
  24. 2024RADIOv2.5: Improved Baselines for Agglomerative Vision Foundation Modelscited by 88
  25. 2023SILC: Improving Vision Language Pretraining with Self-Distillationcited by 66
  26. 2023NAVI: Category-Agnostic Image Collections with High-Quality 3D Shape and Pose Annotationscited by 53
  27. 2022Teaching Matters: Investigating the Role of Supervision in Vision Transformerscited by 45
  28. 2025DUNE: Distilling a Universal Encoder from Heterogeneous 2D and 3D Teacherscited by 28
  29. 2024LiFT: A Surprisingly Simple Lightweight Feature Transform for Dense ViT Descriptorscited by 27
  30. 2024UNIC: Universal Classification Models via Multi-teacher Distillationcited by 24
  31. 2025Learning Task-Agnostic Representations through Multi-Teacher Distillationcited by 2
  32. 2025Fantastic Features and Where to Find Them: A Probing Method to combine Features from Multiple Foundation Modelscited by 2

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2026: 6.

6 papers.

  1. 2026Improved Baselines with Representation Autoencoderscited by 24
  2. 2026V-RAE: Rethinking Video Latent Spaces for Generationcited by 3
  3. 2026LookWhen? Fast Video Recognition by Learning When, Where, and What to Computecited by 1
  4. 2026RayViT: Ray-Conditioned Visual Representations for Viewpoint-Robust Imitation Learningcited by 0
  5. 2026Vision Foundation Models with Synthetic-Only Training for Monocular Spacecraft Pose Estimationcited by 0
  6. 2026IronViT: Toward Efficient Generalist Visual Representation Learningcited by 0