Research map / Video / Action recognition
RNN Fisher Vectors for Action Recognition and Image Annotation
Guy Lev, Gil Sadeh, Benjamin Klein, Lior Wolf
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| use RNN-FV pooling instead of mean pooling | VGG PCA features · HMDB51 | accuracy: 42.16 -> 45.62 | Improved |
| use RNN-FV pooling instead of GMM-FV | VGG PCA features · HMDB51 | accuracy: 36.8 -> 45.62 | Improved |
| use RNN-FV pooling instead of GMM-FV | C3D features · UCF101 | accuracy: 80.04 -> 82.33 | Improved |
| project VGG frames with image-sentence CCA instead of PCA | RNN-FV · UCF101 | accuracy: 79.29 -> 79.49 | Improved |
| fuse RNN-FV of C3D and VGG-CCA with idt | RNN-FV · HMDB51 | accuracy: 54.33 -> 67.71 | Improved |
| fuse RNN-FV of C3D and VGG-CCA with idt | RNN-FV · UCF101 | accuracy: 88.01 -> 94.08 | Improved |
| use regression RNN instead of classification RNN (ablation) | RNN-FV w2v · Flickr8K | sentence mean rank: 13.2 -> 10.4 | Improved |
| use CCA word embedding instead of word2vec (ablation) | RNN-FV · Flickr8K image annotation | recall@1: 27.4 -> 30.9 | Improved |
| train RNN on dataset sentences instead of Wikipedia (ablation) | RNN-FV · Flickr8K image annotation | recall@1: 29.3 -> 27.4 | No clear effect |
| add GMM+HGLMM model to the CCA RNN-FV ensemble | RNN-FV cca · Flickr8K image annotation | recall@1: 30.9 -> 32.9 | Improved |
| add reversed-sentence model to the CCA RNN-FV (ablation) | RNN-FV cca · Flickr8K image annotation | recall@1: 30.9 -> 30.8 | No clear effect |
| reduce gradient dimension to 4,096 with PCA (ablation) | RNN-FV · Flickr8K/Flickr30K/COCO | recall | No clear effect |
Closest papers by meaning
- Making Convolutional Networks Recurrent for Visual Sequence Learning (CVPR 2018 · similarity 0.87)
- Learning Compact Recurrent Neural Networks With Block-Term Tensor Decomposition (CVPR 2018 · similarity 0.87)
- Adaptive RNN Tree for Large-Scale Human Action Recognition (ICCV 2017 · similarity 0.87)
- RPAN: An End-To-End Recurrent Pose-Attention Network for Action Recognition in Videos (ICCV 2017 · similarity 0.86)
- Memory-Augmented Temporal Dynamic Learning for Action Recognition (AAAI 2019 · similarity 0.86)
- A Hybrid RNN-HMM Approach for Weakly Supervised Temporal Action Segmentation (TPAMI 2018 · similarity 0.85)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
