Implementation of Action Recognition using 3D Convnet on UCF-101 dataset.
A repository with a basic layer of 3D deformable receptive field for 3D VoxCNN and 3D VoxResNet
Classification of hand tracks into actions using ResNet models, LSTM or MF-Net. Experiments targeting Epic Kitchens
Video-based surgical skill assessment using 3D convolutional neural networks
Video classification project using 3DCNNs and multiple self-supervised pretraining tasks (June, 2018).
Convert full-width SBS (Side-by-Side) 3D videos into red/cyan anaglyph with automatic cross-correlation focus alignment. Multi-threaded CPU pipeline, optional CUDA acceleration, and customizable FPS & color modes. Simple CLI prompts make it easy to generate aligned anaglyph output fast.
To associate your repository with the 3d-convnet topic, visit your repo's landing page and select "manage topics."