Activity Recognition
Human Activity Recognition is the problem of identifying events performed by humans given a video input. It is formulated as a binary (or multiclass) classification problem of outputting activity class labels. Activity Recognition is an important problem with many societal applications including smart surveillance, video search/retrieval, intelligent robots, and other monitoring systems.
Source: Learning Latent Sub-events in Activity Videos Using Temporal Attention Filters
Papers
Showing 961–970 of 1322 papers
Benchmark Results
| # | Model | Metric | Claimed | Verified | Status |
|---|---|---|---|---|---|
| 1 | Structured Keypoint Pooling | Accuracy | 93.4 | — | Unverified |
| 2 | Semi-Supervised Hard Attention (SSHA); pretrained on Deepmind Kinetics dataset | Accuracy | 90.4 | — | Unverified |
| 3 | Human Skeletons + Change Detection | Accuracy | 90.25 | — | Unverified |
| 4 | Separable Convolutional LSTM | Accuracy | 89.75 | — | Unverified |
| 5 | SPIL Convolution | Accuracy | 89.3 | — | Unverified |
| 6 | Flow Gated Network | Accuracy | 87.25 | — | Unverified |
| # | Model | Metric | Claimed | Verified | Status |
|---|---|---|---|---|---|
| 1 | Boutaleb et al. | 1:1 Accuracy | 97.91 | — | Unverified |
| # | Model | Metric | Claimed | Verified | Status |
|---|---|---|---|---|---|
| 1 | all-landmark-model | Activity Recognition | 0.76 | — | Unverified |