this is the release code for Transferring Object-Scene Convolutional Neural Networks for Event Recognition in Still Images:
Transferring Object-Scene Convolutional Neural Networks for Event Recognition in Still Images
Limin Wang, Zhe Wang, Yu Qiao and Luc Van Gool, http://arxiv.org/abs/1609.00162
The performance is as below:
ChaLearn | WIDER | UIUC | |
---|---|---|---|
acc: | 87.1 | 53.0 | 98.8 |
- feature_extraction_exmaple.m -- you can use this code as template to extract feature as feed to SVM/softmax to get your own score
- prepare_image.m -- serve as function to pre-process image patches
1. Download code and model
2. Extract features (feature_extraction_example.m and prepare_image.m)
3. Feed the feature to SVM/softmax to get the score.
- Code and Files needed to plot figures below lies in picture folder