Free Information Center
KPConv (kernel point convolution) is a convolutional operation designed specifically for deep learning on 3D point clouds. It enables the extraction of geometric features directly from irregular point data, improving tasks such as 3D shape understanding and segmentation.
GeneFace is a neural network-based system designed to generate realistic talking head videos that generalize across different identities. It uses deep learning techniques to synthesize facial movements and expressions from audio or other inputs, enabling applications in virtual avatars and video synthesis.
cuDNN is a GPU-accelerated library developed by NVIDIA that provides optimized routines for deep neural networks. It is widely used to improve the performance of training and inference in deep learning frameworks.
A Message Passing Neural Network (MPNN) is a type of graph neural network designed to process graph-structured data by iteratively exchanging and aggregating information between nodes. It is widely used in fields such as chemistry, social network analysis, and recommendation systems.
ESPnet is an open-source toolkit designed for end-to-end speech processing, encompassing speech recognition, speech synthesis, and related tasks. It provides a unified platform leveraging deep learning architectures to streamline development and research in speech technologies.
DeepLabCut is an open-source software tool for animal pose estimation using deep learning techniques.
DeepSpeech is an open-source speech-to-text engine developed by Mozilla that uses deep learning techniques to convert spoken language into written text. It is designed to enable efficient and accurate automatic speech recognition (ASR) accessible to developers and researchers.
Mila is a research institute specializing in artificial intelligence based in Montreal, Canada. It focuses on machine learning and deep learning research and collaborates with academic and industry partners worldwide.
Linformer is a neural network architecture designed to reduce the computational complexity of the Transformer model’s self-attention mechanism from quadratic to linear with respect to the input sequence length. It achieves this by projecting the attention computation into a lower-dimensional space, enabling efficient processing of long sequences.
Neural voice cloning is a technology that uses deep learning to replicate a person’s voice by generating synthetic speech that closely mimics the original speaker’s vocal characteristics. It enables the creation of personalized speech synthesis with relatively small amounts of audio data.