AI Paper Atlas

Research map / Benchmarks / K400

K400 leaderboard

11 papers at ICML, NeurIPS, ICLR, CVPR, AAAI, ICCV and ECCV, and the journals TPAMI, IJCV and JMLR report results on K400. Best accuracy: 87.7 (V-JEPA 2.1 ViT-G, ECCV 2026).

Numbers are as each paper reports them, read by AI from its results tables and not checked by hand. Papers can differ in setting (training data, model size, extra data); the topic column shows what each paper is about. Open the paper before you compare or quote a number.

K400 · accuracy

11 papers, each with its best reported number for its own method; higher is better.

#accuracy ↑MethodPaperTopic
187.7V-JEPA 2.1 ViT-GV-JEPA 2.1: Unlocking Dense Features in Video Self-Supervised Learning
ECCV 2026 · Table 6
Action recognition
287.2GenRec
Stable Video Diffusion
GenRec: Unifying Video Generation and Recognition with Diffusion Models
NeurIPS 2024 · Main Results table
Video generation
385.3CASTCAST: Cross-Attention in Space and Time for Video Action Recognition
NeurIPS 2023 · Table 3 (main comparison)
Action recognition
484.1M²-CLIP-B/16 (32 frames)
CLIP ViT-B/16
A Multimodal, Multi-Task Adapting Framework for Video Action Recognition
AAAI 2024 · Table: Performance comparison
CLIP and zero-shot
582.2AMD-B, 800 epochs
ViT-B
Asymmetric Masked Distillation for Pre-Training Small Foundation Models
CVPR 2024 · Table 5
Knowledge distillation
682.1EVEREST
ViT-L
EVEREST: Efficient Masked Video Autoencoder by Removing Redundant Spatiotemporal Tokens
ICML 2024 · Table: Performance Comparison
Video segmentation and tracking
781.1BEVT†
Video Swin-Base
BEVT: BERT Pretraining of Video Transformers
CVPR 2022 · Table: Comparison to state-of-
Video segmentation and tracking
877.6TSM Taylor
TSM, trained from scratch
Taylor Videos for Action Recognition
ICML 2024 · Table 3
Action recognition
975.43DyT† N=4
ViT-B/16
Dynamic Tuning Towards Parameter and Inference Efficiency for ViT Adaptation
NeurIPS 2024 · Effectiveness of MoE-Adapter t
Pruning and efficient tuning
1070ρBYOL, R-50, 8 frames
R-50
A Large-Scale Study on Unsupervised Spatiotemporal Representation Learning
CVPR 2021 · Table (Comparison with state-o
Video segmentation and tracking
1 more papers on this leaderboard, the baselines reported on it and a CSV download are free with a scime account.
Sign in free to see all

Not affiliated with any conference, journal or the benchmark's authors. Corrections: contact@latenticlabs.com.