30 dépôts
Standardized training routines and scripts for image-based neural network architectures.
Distinguishing note: Focuses on image-specific model training workflows rather than general-purpose NLP or tabular data training.
Explore 30 awesome GitHub repositories matching artificial intelligence & ml · Computer Vision Training. Refine with filters or upvote what's useful.
DeepSpeed is a high-performance library designed to scale deep learning model training and inference across massive clusters of GPUs and compute nodes. It provides a comprehensive suite of tools for distributed training, enabling the execution of models that exceed the memory capacity of single devices through advanced parameter partitioning, pipeline-based model parallelism, and memory-efficient state offloading. The framework distinguishes itself through specialized communication-efficient optimizers and hardware-aware acceleration techniques. By utilizing gradient compression, quantization
The framework includes standard training scripts for image datasets to verify model performance and establish baseline accuracy metrics for neural networks.
This project is a library of pretrained computer vision architectures and backbones for image classification and feature extraction. It serves as a comprehensive model zoo and collection of standardized image encoders, including ResNet, Vision Transformers, and EfficientNet, for use in visual analysis and as backbones for object detection and image segmentation. The library provides a framework for distributed training and evaluation of image models using advanced data augmentation and optimization scripts. It includes a dedicated toolset for converting trained PyTorch vision models into the
Provides optimized training routines and scripts specifically for image-based neural network architectures.
This project is a comprehensive library of state-of-the-art neural network architectures designed for image classification and feature extraction. It provides a complete deep learning training framework that supports distributed execution, allowing users to build, train, and fine-tune vision models using optimized schedulers and pre-configured training recipes. The library distinguishes itself through a modular backbone architecture that treats neural networks as decoupled feature extractors, enabling the retrieval of multi-scale outputs for downstream tasks like object detection and segmenta
Building and training deep learning models for image classification by leveraging distributed training scripts, optimized schedulers, and pre-configured training recipes.
Swin-Transformer is a deep learning framework designed for training and deploying hierarchical vision transformer models. It serves as a research library and toolkit for computer vision tasks, providing the infrastructure to build models that replace standard convolution operations with sliding window self-attention mechanisms. By utilizing a multi-scale feature hierarchy, the framework enables the processing of visual data at varying resolutions and spatial scales. The project distinguishes itself through its implementation of shifted window partitioning, which facilitates global information
Provides standardized training routines for transformer-based computer vision models on large-scale infrastructure.
PaddleDetection is an object detection framework designed for the end-to-end development, training, and deployment of computer vision models. It provides a comprehensive library of modular neural network architectures and pipelines that support object detection, instance segmentation, and multi-object tracking tasks. The project distinguishes itself through a configuration-driven approach that decouples model components like backbones and heads, allowing for the flexible assembly of custom vision workflows. It incorporates advanced techniques such as anchor-free detection logic, joint detecti
Provides comprehensive training routines for object detection, segmentation, and tracking models.
This project is a machine learning data pipeline designed to automate the collection, curation, and preparation of large-scale image datasets. It functions as an image dataset scraper and computer vision curator, providing the necessary infrastructure to aggregate categorized files from web sources and organize them into structured directories for model development. The system distinguishes itself through a batch-processing architecture that integrates data acquisition with automated integrity validation. By scanning files to remove corrupted or invalid images and applying deterministic parti
Provides standardized training routines for preparing and validating image-based neural network models.
This project is a collection of educational resources and implementation frameworks providing deep learning model recipes, code samples, and step-by-step guides for computer vision tasks. It organizes complex workflows into modular recipes and implementation guides to facilitate the building of image and video analysis models. The framework focuses on specialized vision capabilities, including an image similarity framework for fast retrieval and re-ranking, human pose estimation, and video action recognition. It also provides specific tools for crowd density estimation and document image clea
Trains deep neural networks to compute image representations for identifying similar images.
Keras-GAN is a collection of generative adversarial network implementations built with Keras for synthetic data generation and image manipulation. It provides frameworks for image-to-image translation, image inpainting, and neural image super-resolution. The library includes tools for learning disentangled latent space representations to control specific attributes of synthetic outputs. It also features capabilities for image domain translation using paired or unpaired data and the ability to fill corrupted or missing image parts by analyzing surrounding visual context. The project covers ge
Extracts structured and disentangled representations to allow precise control over specific attributes of synthetic outputs.
ImageAI is a Python computer vision library providing a suite of tools for image classification, object detection, and video analytics. It functions as an integrated framework for locating and labeling objects in static images and video streams, utilizing deep learning models for identification and categorization. The project includes a model training toolkit that allows for the creation of custom classifiers and detectors through scratch training or transfer learning. It features a GPU-accelerated inference engine to increase processing speed for vision tasks and includes specialized utiliti
Prepares image datasets and trains detection networks from scratch or via transfer learning.
DAIN is a video frame synthesis engine and AI video upsampling tool designed to increase video playback smoothness. It functions as a computer vision model that synthesizes intermediate frames between existing images to transform low frame rate video into high frame rate content. The system utilizes depth-aware video frame interpolation to predict the motion of pixels between consecutive images. By analyzing spatial depth via depth maps, the tool generates new frames that account for occlusions and overlapping objects to create slow motion effects. The framework incorporates optical flow int
Utilizes standardized training routines on triplet datasets to improve synthesized frame accuracy.
Infinigen is a procedural 3D scene generation framework that creates photorealistic indoor and outdoor environments for computer vision training data. It combines constraint-based object placement, GPU geometry shaders, and ground-truth rendering passes to produce scenes with depth, normals, and segmentation masks alongside final images. The framework distinguishes itself through modular asset composition, a node-graph material system, and physics simulation integration that embeds rigid-body and fluid dynamics directly into the generation pipeline. Procedural rule-based scene composition and
Produces photorealistic scenes with ground-truth annotations to support supervised model training.
Corenet is a deep learning training framework and computer vision model library designed for developing neural networks across vision, text, and audio modalities. It functions as a distributed training orchestrator for scaling workloads across multiple compute nodes and provides a multimodal data pipeline for processing image, text, and video data. The project includes a model conversion toolkit for transforming weights and architectures between different machine learning frameworks. It also provides tools for optimizing model performance on Apple Silicon and reducing response latency in gene
Provides standardized training routines and scripts for image-based neural network architectures.
EasyPR is an automatic license plate recognition system designed to detect vehicle license plates and extract alphanumeric characters from images of Chinese vehicles. It functions as a deep learning OCR tool that converts image regions of license plates into machine-readable text strings. The system includes a specialized detector for identifying vehicle plates within unconstrained environments and complex visual backgrounds. It also provides a synthetic data generator to create artificial image datasets used to train and improve the accuracy of the recognition models. The project covers a m
Provides training routines for neural networks to improve the accuracy of character and plate detection.
This project is a collection of TensorFlow 2.x machine learning tutorials and practical code examples. It serves as a deep learning implementation guide for constructing diverse neural network architectures, including convolutional, recurrent, and generative networks. The repository provides templates and examples for several specialized domains, including computer vision for image classification and object detection, natural language processing for text generation and language understanding, and generative AI for synthesizing data using adversarial networks and autoencoders. It also includes
Provides standardized training routines and scripts for image-based neural network architectures.
Ce projet est un framework de traitement du langage naturel axé sur un pré-entraîneur autorégressif généralisé conçu pour la représentation de langage non supervisée. Il implémente un modèle de langage qui combine un entraînement basé sur la permutation avec une architecture Transformer-XL pour fonctionner comme un processeur de texte à long contexte. Le système se distingue par sa capacité à gérer des séquences de texte dépassant les limites de longueur standard grâce à l'utilisation de la récurrence au niveau du segment et du codage positionnel relatif. Il met à l'échelle le pré-entraînement haute performance sur plusieurs clusters de GPU et TPU en utilisant des implémentations d'entraînement distribué. La base de code couvre l'ensemble du workflow de machine learning, incluant le nettoyage de texte et la tokenisation par sous-mots pour le prétraitement des données, ainsi que le fine-tuning spécifique à une tâche pour la réponse aux questions, la compréhension de lecture et la classification de texte. Il inclut des utilitaires pour l'optimisation des paramètres, la planification du taux d'apprentissage et l'évaluation des probabilités de réponse via des métriques de précision-rappel. Le projet fournit des configurations pour gérer les hyperparamètres du modèle et l'entraînement accéléré par le matériel sur plusieurs hôtes.
Learns general semantic and syntactic patterns from unlabeled text for use in downstream tasks.
Ce projet est une collection de Jupyter Notebooks éducatifs proposant des tutoriels sur la construction de réseaux de neurones et les opérations sur tenseurs avec le framework TensorFlow. Il sert de dépôt pédagogique pour le machine learning et de guide d'implémentation pour les étudiants en deep learning. La suite se concentre sur des architectures avancées spécifiques, notamment les réseaux convolutifs pour la classification d'images, les réseaux résiduels avec connexions sautées pour la stabilité de l'entraînement, et les auto-encodeurs variationnels pour la modélisation générative et la synthèse de données. Elle inclut également des guides pour construire des auto-encodeurs de débruitage et profonds afin d'effectuer l'extraction de caractéristiques et la réduction de dimensionnalité. Le dépôt couvre un large spectre de modélisation prédictive, avec des implémentations de régression linéaire, polynomiale et logistique pour prédire des valeurs continues et des résultats binaires. Le contenu est organisé en notebooks interactifs permettant aux utilisateurs d'exécuter des opérations mathématiques et de modifier des expériences de machine learning.
Trains neural networks to learn compressed representation embeddings using Gaussian noise in variational autoencoders.
Neuraltalk2 est un système de vision par apprentissage profond conçu pour la génération automatique de légendes d'images. Construit avec PyTorch, il utilise une architecture hybride qui combine un encodeur de réseau neuronal convolutif avec un décodeur de réseau neuronal récurrent pour générer des descriptions textuelles à partir d'entrées visuelles. Le projet propose un pipeline d'entraînement accéléré par GPU capable de distribuer les charges de travail sur plusieurs unités de traitement graphique via une distribution multi-processus. Il prend en charge la génération de descriptions pour des fichiers d'images statiques et des flux vidéo en temps réel. Le framework inclut des capacités pour le réglage fin de l'encodeur, l'échantillonnage de texte par recherche en faisceau (beam search) avec contrôle de température, et l'utilisation de métriques linguistiques standard de l'industrie pour évaluer la précision et la fluidité des légendes. Il fournit également des utilitaires pour le prétraitement des jeux de données, la persistance des points de contrôle du modèle et l'exportation des prédictions dans des fichiers JSON structurés. L'implémentation est fournie sous forme de Jupyter Notebook.
Provides standardized training routines to associate visual features with natural language descriptions.
Super-Gradients est un framework de vision par ordinateur PyTorch et une bibliothèque d'entraînement conçue pour le cycle de vie complet des modèles de vision. Il fonctionne comme un optimiseur de modèle de deep learning et une boîte à outils de déploiement pour entraîner et affiner des modèles à travers des tâches de classification d'image, détection d'objets, segmentation sémantique et estimation de pose. Le projet fournit des outils spécifiques pour l'optimisation de modèle, incluant la distillation de connaissances enseignant-élève et la compression de précision numérique pour réduire les besoins en mémoire et en calcul. Il inclut également l'implémentation de l'architecture Yolo-NAS pour la détection d'objets haute performance. Le framework couvre une large surface de capacités incluant l'entraînement GPU distribué, les pipelines de vision modulaires et l'automatisation des runs d'entraînement via des configurations de recettes structurées. Il gère en outre le chargement des données, l'augmentation d'image et l'exportation des poids entraînés dans des formats universels pour les accélérateurs matériels de production.
Provides standardized training routines and scripts for image-based neural network architectures across multiple vision tasks.
Co-tracker is a PyTorch point tracking framework and dense point tracking model designed to map the motion of individual pixels throughout a video. It functions as a video pixel tracker that predicts point trajectories and visibility masks across sequences of video frames. The project includes a computer vision training pipeline that utilizes teacher-student knowledge distillation. This allows for the generation of pseudo-labels from unannotated real video data to fine-tune pre-trained models and reduce the gap between synthetic and real data environments. The framework provides capabilities
Provides a training pipeline for fine-tuning point tracking models using teacher-student distillation.
Neuralangelo is a neural surface reconstruction framework that transforms two-dimensional image sequences and multi-view photography into high-fidelity 3D meshes. It implements a pipeline for training neural radiance fields to represent complex scenes as digital geometry. The project utilizes a signed distance function for surface representation and multi-resolution hash encoding to capture both coarse and fine geometric details. It employs differentiable volume rendering and gradient-based eikonal regularization to ensure the learned distance functions remain physically plausible. The syste
Provides a GPU-accelerated training workflow for processing image datasets and estimating camera poses.