Dynamic Scene Recognition on YUPENN (leave-one-out)
98.1AccuracyC3D
Evaluation Results
| Method | Links | |
|---|---|---|
| C3Dclassifier=linear SVM, clip length=16 frames, aggregation=simple averaging2014.12 | 98.1 | |
| Imagenetfeatures=Imagenet pre-trained baseline2014.12 | 96.7 | |
| [9]feature encoding=FV, LLC, and dynamic pooling2014.12 | 96.2 | |
| MAS (Ours)Backbone=C3D, Pre-training=Self-supervised (MAS), Classifier=Linear SVM, Evaluation Protocol=leave-one-out2019.04 | 90.2 | |
| [12]protocol=leave-one-out2019.04 | 86.9 | |
| [8]2014.12 | 86 | |
| [10]protocol=leave-one-out2019.04 | 86 | |
| [41]2014.12 | 85 | |
| [4]2014.12 | 80.7 | |
| [8]protocol=leave-one-out2019.04 | 80.7 | |
| [27]protocol=leave-one-out2019.04 | 76.67 | |
| [43]protocol=leave-one-out2019.04 | 70.47 |