Please i need an extracted video datasets using deep learning or any video feature extraction method using deep learning for my current project.
https://ieeexplore.ieee.org/document/7862078/
https://www.researchgate.net/post/do_we_need_any_feature_extraction_of_image_to_train_Deep_neural_network_for_image_classifications
https://www.researchgate.net/post/How_to_store_vectors_of_features_extracted_from_video_frames_to_perform_video_classification
https://www.researchgate.net/post/What_is_the_best_way_for_motion_feature_extraction_form_video_frames
Hi,
You can use two streams CNN architecture to extract the video features.
The source code of two streams CNN is widely available in GitHub.
Links: Article Two-Stream Convolutional Networks for Action Recognition in Videos
Another solution is to use 3D-CNN.
"C3D: Generic Features for Video Analysis"
Links: https://arxiv.org/abs/1412.0767v1
Hope it helps your research.
I think this link can help you "https://github.com/vra/action-recognition-using-3d-resnet". It provides extracted features from UCF101 and HMDB51 datasets.
As a final note, I didn't understand fully your question, do you need extracted features or a pipeline to extract features from videos (using CNNs)?
Dr Arthur thanks so much for your recommendations. i wanted either of them or both for video.
- Badges
- Science topic
- Similar topics
- Phytochemistry
- Extracts