Action Recognition In Videos On Hmdb 51

Metrics

Average accuracy of 3 splits

Results

Performance results of various models on this benchmark

Comparison Table

Model Name	Average accuracy of 3 splits
action-recognition-with-trajectory-pooled	65.9
end-to-end-learning-of-motion-representation	72.6
late-temporal-modeling-in-3d-cnn	85.10
contextual-action-cues-from-camera-sensor-for	80.92
learning-discriminative-video-representations	74.3
actionflownet-learning-motion-representation	56.4
appearance-and-relation-networks-for-video	70.9
vimpac-video-pre-training-via-masked-token	65.9
two-stream-convolutional-networks-for-action	59.4
rethinking-spatiotemporal-feature-learning	75.9
a-closer-look-at-spatiotemporal-convolutions	76.4
videomoco-contrastive-video-representation	49.2
high-order-tensor-pooling-with-attention-for	85.70
quo-vadis-action-recognition-a-new-model-and	77.3
representation-flow-for-action-recognition	81.1
faster-recurrent-networks-for-video	75.7
a-closer-look-at-spatiotemporal-convolutions	66.6
dynamic-image-networks-for-action-recognition	65.2
convnet-architecture-search-for	54.9
paying-more-attention-to-motion-attention	72.0
videomoco-contrastive-video-representation	43.6
quo-vadis-action-recognition-a-new-model-and	80.9
d3d-distilled-3d-networks-for-video-action	80.5
omni-sourced-webly-supervised-learning-for	83.8
quo-vadis-action-recognition-a-new-model-and	80.7
learning-spatio-temporal-representation-with-3	78.9
d3d-distilled-3d-networks-for-video-action	78.7
temporal-segment-networks-towards-good	69.4
dmc-net-generating-discriminative-motion-cues	62.8
hidden-two-stream-convolutional-networks-for	78.7
quo-vadis-action-recognition-a-new-model-and	74.3
distinit-learning-video-representations	54.8
quo-vadis-action-recognition-a-new-model-and	77.1
pose-and-joint-aware-action-recognition	54.2
learning-spatiotemporal-features-with-3d	51.6
convolutional-two-stream-network-fusion-for	65.4
pose-and-joint-aware-action-recognition	84.53
spatiotemporal-residual-networks-for-video	70.3
r-stan-residual-spatial-temporal-attention	55.16
self-supervised-video-transformer	67.2
r-stan-residual-spatial-temporal-attention	62.8
holistic-large-scale-video-understanding	76.5
optical-flow-guided-feature-a-fast-and-robust	74.2
learning-spatio-temporal-representations-with	71.5
a-closer-look-at-spatiotemporal-convolutions	78.7
dmc-net-generating-discriminative-motion-cues	71.8
smart-frame-selection-for-action-recognition	84.36
tensor-representations-for-action-recognition	86.11
d3d-distilled-3d-networks-for-video-action	79.3
a-closer-look-at-spatiotemporal-convolutions	74.5
learning-spatio-temporal-representation-with-3	75.7
hierarchical-feature-aggregation-networks-for	71.13
spatiotemporal-multiplier-networks-for-video	72.2
bubblenet-a-disperse-recurrent-structure-to	82.60
motionsqueeze-neural-motion-feature-learning	77.4
a-closer-look-at-spatiotemporal-convolutions	72.7
dmc-net-generating-discriminative-motion-cues	77.8
hallucinating-statistical-moment-and-subspace	87.56
cooperative-cross-stream-network-for	81.9
hallucinating-bag-of-words-and-fisher-vector	82.48
video-classification-with-finecoarse-networks	77.6
ts-lstm-and-temporal-inception-exploiting	69
videomae-v2-scaling-video-masked-autoencoders	88.1
vidtr-video-transformer-without-convolutions	74.4
susinet-see-understand-and-summarize-it	62.7
learning-spatio-temporal-representation-with-3	80.5
a-closer-look-at-spatiotemporal-convolutions	70.1
mars-motion-augmented-rgb-stream-for-action	80.9
quo-vadis-action-recognition-a-new-model-and	74.8
perf-net-pose-empowered-rgb-flow-net	83.2
long-term-temporal-convolutions-for-action	64.8
bidirectional-cross-modal-knowledge	83.1
towards-universal-representation-for-unseen	51.8
asymmetric-masked-distillation-for-pre	79.6
zeroi2v-zero-cost-adaptation-of-pre-trained	83.4
high-order-tensor-pooling-with-attention-for	87.21