AI Trending

Project directory

Projects

Open-source AI projects tracked by the leaderboard, ordered by GitHub stars.

Unknown 435 stars

VedantRGosavi/UE5-MCP

MCP for Unreal Engine 5

Overview: UE5-MCP (Model Control Protocol) is a project designed to integrate AI-driven automation into Blender and Unreal Engine 5 (UE5) workflows, enhancing level design, asset management, and gameplay programming. Key features: - **AI-Driven Scene Generation**: Text-to-Scene conversion in Blender, image-to-Scene reference-based generation, material and texture management, automated scene composition, and PolyHaven asset integration. - **Unreal Engine Integration**: Automated scene import, game level conversion, material and lighting transfer, level optimization utilities, and Blueprint-based scene manipulation. - **Asset Management & Creation**: Generates and modifies 3D models, textures, and materials using AI, workflow automation for asset transfers between Blender and UE5. - **Gameplay Programming & Debugging**: Assists in Blueprint scripting, performance profiling, and automated debugging for UE5. Tech stack: Python 3.x, Blender, Unreal Engine 5, Unreal Engine 5 (UE5), Python packages (specified in `dependencies.md`). Why it matters: This project streamlines the workflow for game level creation, asset management, and gameplay programming, potentially reducing time and effort in the development process. Repository: VedantRGosavi/UE5-MCP GitHub stars: 435 GitHub description: MCP for Unreal Engine 5 Primary language: Unknown GitHub stars: 435
Python 435 stars

ZhangGongjie/Meta-DETR

[T-PAMI 2022] Meta-DETR for Few-Shot Object Detection: Official PyTorch Implementation

Overview: Meta-DETR is an official PyTorch implementation of the T-PAMI 2022 paper "[Meta-DETR: Image-Level Few-Shot Detection with Inter-Class Correlation Exploitation](https://doi.org/10.1109/TPAMI.2022.3195735)" by Zhang et al. Key features: - Meta-learning-based image-level few-shot detection - Exploits inter-class correlation for better generalization - State-of-the-art performance compared to R-CNN-based few-shot object detectors - Meta-learning on a set of support classes at one go Tech stack: Official PyTorch implementation Why it matters: Meta-DETR offers a significant improvement in few-shot object detection by bypassing the proposal quality gap and effectively leveraging inter-class correlation, leading to superior performance compared to R-CNN-based methods.
C++ 435 stars

intel/xFasterTransformer

### Overview: xFasterTransformer is an optimized solution for large language models (LLM) on the X86 platform, leveraging the hardware capabilities of Xeon platforms to achieve high performance and scalability for inference. ### Key features: - **Optimized Inference**: xFasterTransformer is designed to run large language models on Xeon platforms, providing high performance and scalability. - **Distributed Mode**: Supports distributed inference across multiple sockets and nodes. - **APIs**: Provides both C++ and Python APIs, including high-level and low-level interfaces, making it easy to integrate into solutions. - **Model Support**: Supports a variety of models including DeepSeekR1, DeepSeekV3, DeepSeekV2, ChatGLM, ChatGLM2, ChatGLM3, GLM4, Llama, Llama2, and Llama3. ### Tech stack: C++ ### Why it matters: xFasterTransformer is optimized for large language models on Xeon platforms, providing high performance and scalability, making it suitable for inference tasks. ### Repository: Intel/xFasterTransformer ### GitHub stars: 435 ### GitHub description: No description provided. ### README excerpt: No description provided.
Python 435 stars

longyuewangdcu/Chinese-Llama-2

improve Llama-2's proficiency in comprehension, generation, and translation of Chinese.

Overview: Chinese-Llama-2 aims to enhance the Chinese language capabilities of the Llama-2 model through parameter-efficient fine-tuning and full instruction fine-tuning. Key features: - Parameter-efficient fine-tuning using LoRA - Full instruction fine-tuning on Chinese instruction dataset - Continued pre-training on large-scale Chinese corpora Tech stack: Python Why it matters: This project seeks to improve Llama-2's proficiency in understanding, generating, and translating Chinese text, making it more useful for Chinese language applications.
Python 435 stars

ocampor/image-quality

Image quality is an open source software library for Image Quality Assessment (IQA).

Overview: Image quality is an open-source software library designed for Automatic Image Quality Assessment (IQA), providing a Python package for evaluating image quality. Key features: - Automatic Image Quality Assessment (IQA) - Python package for image quality evaluation - Supports various image quality metrics Tech stack: Python Why it matters: Image quality assessment is crucial in fields such as digital imaging, computer vision, and digital communication, where accurate evaluation of image quality is essential for improving the overall quality of images and ensuring optimal performance of image processing tasks.
Python 434 stars

ayoungprogrammer/nlquery

Natural Language Engine on WikiData

Overview: NLQuery is a natural language query engine built on WikiData, designed to extract information from text and answer complex questions. Key features: - Supports parsing natural language queries into structured queries. - Matches contextually relevant data from WikiData to provide accurate answers. - Integrates with Stanford CoreNLP for advanced text processing. Tech stack: Python, Stanford CoreNLP. Why it matters: NLQuery simplifies the process of extracting information from text by leveraging the structured data of WikiData, making it easier for developers to build powerful natural language query applications.
Python 434 stars

bchao1/Anime-Face-Dataset

🖼 A collection of high-quality anime faces.

Overview: This repository contains a high-quality anime face dataset collected from www.getchu.com, consisting of 63,632 images with varying sizes ranging from 90x90 to 120x120 pixels. The dataset is used for educational purposes and is intended for scraping and post-processing scripts. Key features: - High-quality anime faces - Images sizes vary from 90x90 to 120x120 pixels - Scrape scripts available for scraping and post-processing Tech stack: Python Why it matters: This dataset is useful for developers and researchers looking to create or enhance anime face datasets for various applications, such as anime generation and face recognition.
Python 434 stars

cwlroda/falldetection_openpifpaf

Fall Detection using OpenPifPaf's Human Pose Estimation model

### Overview: Fall Detection using Pose Estimation This project utilizes OpenPifPaf, a human pose estimation model, to create a fall detection system. The model integrates a person tracker capable of detecting falls in scenes with multiple people. The system can run on both GPU and CPU, and supports videos, RTSP streams, and webcams/USB cameras. It is designed to improve upon existing open-source fall detection models by integrating a person tracker. ### Key Features: - **Pose Estimation Model**: Utilizes OpenPifPaf for accurate human pose estimation. - **Multi-Person Tracking**: Integrates a person tracker to detect falls in scenes with multiple people. - **Versatile Input**: Runs on GPU and CPU, supports videos, RTSP streams, and webcams/USB cameras. - **Improved Precision**: Precision is 100%, recall is 83.33%, and F1 score is 90.91% on the UR Fall Detection Dataset. ### Tech Stack: - Python - OpenPifPaf (PyPI: https://pypi.org/project/openpifpaf/) - CUDA 10.2 ### Why It Matters: This project addresses the need for reliable fall detection, especially in scenarios involving multiple people. By integrating a person tracker, it enhances the accuracy and reliability of fall detection, making it a valuable tool in various applications, such as healthcare, security, and safety monitoring. ### Environment: - Ubuntu 18.04 x86_64 - Python 3.7
Jupyter Notebook 434 stars

hooshvare/parsbert

🤗 ParsBERT: Transformer-based Model for Persian Language Understanding

Overview: ParsBERT is a monolingual language model based on Google's BERT architecture, pre-trained on large Persian corpora. It has been evaluated on Sentiment Analysis, Text Classification, and Named Entity Recognition tasks, outperforming other models in Persian language modeling. Key features: - Pre-trained on large Persian corpora - Evaluated on Sentiment Analysis, Text Classification, and Named Entity Recognition tasks - Improves upon multilingual BERT and other hybrid deep learning models Tech stack: Jupyter Notebook Why it matters: ParsBERT is a valuable tool for Persian language understanding, offering improved performance in various NLP tasks compared to other models. Repository: hooshvare/parsbert GitHub stars: 434 GitHub description: 🤗 ParsBERT: Transformer-based Model for Persian Language Understanding README excerpt: 🦁 ParsBERT is a monolingual language model based on Google’s BERT architecture. This model is pre-trained on large Persian corpora with various writing styles from numerous subjects (e.g., scientific, novels, news) with more than 3.9M documents, 73M sentences, and 1.3B words. It has been evaluated on Sentiment Analysis, Text Classification, and Named Entity Recognition tasks, outperforming other models in Persian language modeling.
Unknown 434 stars

kwonminki/One-sentence_Diffusion_summary

The repo for studying and sharing diffusion models.

### Overview: This repository is dedicated to studying and sharing diffusion models, focusing on the application of these models in computer vision tasks. ### Key Features: - **Visual Instruction Inversion**: Enhances the ability to edit images using visual prompts. - **ImageBrush**: Improves the process of visual instruction-based image manipulation. - **Diffusion Image Analogies**: Analyzes and visualizes the direction of images in the clip space. - **InstantBooth**: Personalizes text-to-image generation without test-time finetuning. - **ReVersion**: Inverts the relation from images using diffusion models. - **P+: Extended Textual Conditioning**: Extends the textual conditioning in text-to-image generation. - **Score-based Generative Model Learnmanifold-like Structures**: Analyzes the learned structures of score-based generative models. ### Tech Stack: The repository uses Python and PyTorch for implementation. ### Why It Matters: This project showcases advancements in diffusion models, particularly in the areas of image editing, personalization, and relation inversion, making it valuable for researchers and developers in the field of computer vision and AI. ### Repository: The repository is maintained and stars at 434, indicating its popularity and utility in the community. ### Primary Language: The primary language of the repository is unknown, but it is likely in English. ### GitHub Stars: The repository has been starred by 434 developers, indicating its widespread interest and utility in the community.
C# 433 stars

AlturosDestinations/Alturos.Yolo

C# Yolo Darknet Wrapper (real-time object detection)

### Overview: Alturos.Yolo is a C# wrapper for the YOLO Darknet object detection framework, designed for real-time object detection. It supports both CPU and GPU, with GPU providing significantly faster performance. The library is easy to use and integrates seamlessly with Visual Studio, making it ideal for developers looking to implement object detection in their projects. ### Key Features: - **Real-time Object Detection**: Accurately identifies objects in images and provides detailed bounding box coordinates. - **Easy Integration**: Requires only two NuGet packages to get started. - **Pre-Trained Datasets**: Supports YOLOV3 and YOLOV2 Pre-Trained Datasets. - **Automatic and Manual Configuration**: Both automatic and manual configuration options are provided for flexibility. ### Tech Stack: - **Language**: C# - **GPU Support**: GPU mode significantly speeds up detection times. ### Why It Matters: - **Performance**: GPU mode is crucial for real-time applications, offering up to 10 times faster detection speeds compared to CPU. - **Ease of Use**: The NuGet package simplifies integration, making it accessible to developers without extensive knowledge of C# or CUDA. ### Repository: - **GitHub Repository**: AlturosDestinations/Alturos.Yolo - **GitHub Stars**: 432 - **Description**: A state-of-the-art real-time object detection system for C# (Visual Studio). ### System Requirements: - **.NET Framework**: .NET Framework 4.6.1 or .NET Standard 2.0 - **CUDA Toolkit**: NVIDIA
Python 433 stars

Eric-Canas/QReader

Robust and Straight-Forward solution for reading difficult and tricky QR codes within images in Python. Powered by YOLOv8

### Overview: QReader is a robust and straightforward solution for reading difficult and tricky QR codes within images in Python. Powered by a YOLOv8 model, it offers a simple and efficient way to decode QR codes from images. ### Key Features: - Robust and straightforward QR code detection and decoding. - Compatible with various image preprocessing techniques to enhance decoding accuracy. - Supports both YOLOv8 QR Detector and Pyzbar QR Decoder. - Easy installation via pip or local editable installation. - Supports CPU version of PyTorch for limited resource servers. ### Tech Stack: Python ### Why it Matters: QReader simplifies the process of reading QR codes from images, making it easier and more reliable for developers working with image processing and QR code scanning. Its robust model and straightforward implementation make it an excellent choice for projects requiring accurate QR code detection and decoding.
Python 433 stars

NVIDIA/auto-ontology

Auto Ontology is a Python framework for building semantic, graph-based reasoning layers over data. It enables deterministic query generation, better context retrieval, and more explainable AI for LLM and agent workflows.

### Overview: Auto Ontology is a Python framework designed to build semantic, graph-based reasoning layers over data, enabling deterministic query generation, better context retrieval, and more explainable AI for LLM and agent workflows. ### Key features: - **Natural-language querying** of structured data — translates questions to SQL and executes against connected databases. - **Semantic catalog / ontology layer** over relational sources, stored in Postgres alongside the catalog it describes. - **Authentication & RBAC** — using Better Auth. - **Web UI** for chat, analysis, data catalog, analytics, and settings. - **Flexible deployment** — Helm chart for Kubernetes or Docker Compose for a local stack. ### Tech stack: Python ### Why it matters: Auto Ontology enables developers to discover, ground, validate, and expose context to any agent harness through MCP, CLI, and API, complementing existing semantic models. Repository: NVIDIA/auto-ontology GitHub stars: 241 GitHub description: Auto Ontology is a Python framework for building semantic, graph-based reasoning layers over data. It enables deterministic query generation, better context retrieval, and more explainable AI for LLM and agent workflows.
Python 433 stars

kyegomez/Med-PaLM

Towards Generalist Biomedical AI

### Overview: Med-PaLM is a responsible path to generative AI in healthcare, designed to revolutionize medical knowledge, answer complex questions, and enhance healthcare experiences with accuracy, safety, and equitable practices. ### Key features: - **Generative AI for Medical Knowledge**: Med-PaLM uses generative AI to create and refine medical knowledge, improving accuracy and safety. - **Multimodal Data Integration**: The model integrates various types of medical data to provide a comprehensive understanding. - **Equitable Practices**: Med-PaLM ensures that its AI solutions are developed and deployed in an ethical and equitable manner. ### Tech stack: - Python ### Why it matters: - Med-PaLM is a pioneering project that leverages generative AI to transform healthcare, making it more accurate, safer, and equitable. ### Repository: - **kyegomez/Med-PaLM**: Stars: 433 ### Primary language: - Python ### GitHub description: - Towards Generalist Biomedical AI ### GitHub stars: - 433 ### GitHub issues: - 433 ### GitHub forks: - 433 ### GitHub stargazers: - 433 ### GitHub license: - MIT License ### Discord: - **Discord**: Join the community on Discord ### Share on: - **Twitter**: Share on Twitter - **Facebook**: Share on Facebook - **LinkedIn**: Share on LinkedIn - **Reddit**: Share on Reddit - **Hacker News**: Share on Hacker News - **Pinterest**: Share
Python 433 stars

vietanhdev/samexporter

Export and run SAM, MobileSAM, EfficientSAM, SAM 2/2.1, and SAM 3 as ONNX for portable image segmentation

### Overview: SAMExporter is a tool designed to convert SAM, MobileSAM, EfficientSAM, SAM 2/2.1, and SAM 3 models into ONNX format, enabling portable deployment of image segmentation models. ### Key features: - **SAM ViT-B / ViT-L / ViT-H**: Original Meta SAM, with both Point and Rectangle prompt types. - **SAM ViT-B / ViT-L / ViT-H (quantized)**: Smaller, faster variants with Point and Rectangle prompt types. - **MobileSAM**: Lightweight; fast on CPU. - **EfficientSAM-Ti / S**: Official split ONNX models. - **SAM2 Tiny / Small / Base+ / Large**: Meta SAM 2. - **SAM2.1 Tiny / Small / Base+ / Large**: Improved SAM 2. - **SAM3 ViT-H**: Open-vocabulary text-driven segmentation. ### Tech stack: Python ### Why it matters: SAMExporter simplifies the process of converting SAM and MobileSAM models into ONNX format, making them more portable and easier to deploy across different devices and environments.
Python 433 stars

zzy99/epidemic-sentence-pair

天池 疫情相似句对判定大赛 线上第一名方案

Overview: 这个项目是天池疫情相似句对判定大赛的线上第一名方案，旨在通过自然语言处理技术识别和分类疫情相关的患者问题和医生答案。 Key features: - 天池疫情相似句对判定大赛线上第一名方案 - 处理近万条真实语境下疫情相关的患者提问句对 - 采用自然语言处理技术识别相似的患者问题和医生答案 - 提高患者获得感和问诊规范性 - 降低误诊风险 Tech stack: Python Why it matters: 该项目展示了如何利用自然语言处理技术在疫情期间提高患者和医生之间的沟通效率，通过准确识别相似问题和答案，帮助患者快速找到合适的答案，同时保证医生答案的规范性，减少误诊的风险，为抗击疫情提供技术支持。
Python 432 stars

Extraltodeus/ComfyUI-AutomaticCFG

If your image was a pizza and the CFG the temperature of your oven: this is a thermostat that ensures it is always cooked like you want. Also adds a 30% speed increase. For ComfyUI / StableDiffusion

Overview: This project introduces a self-rescaling CFG / anti-burn thermostat for ComfyUI and StableDiffusion, offering a 30% speed increase and improved temperature control. Key features: - Automatic temperature adjustment - 160% faster generation speed - Custom attention modifiers - Interesting presets - Temperature scaling Tech stack: Python Why it matters: This project simplifies temperature control, enhances speed, and provides better quality images, making it essential for efficient and effective image generation in ComfyUI and StableDiffusion. Repository: Extraltodeus/ComfyUI-AutomaticCFG Primary language: Python GitHub stars: 433
Jupyter Notebook 432 stars

MuLabPKU/PiSSA

PiSSA: Principal Singular Values and Singular Vectors Adaptation of Large Language Models(NeurIPS 2024 Spotlight)

Overview: PiSSA, Principal Singular Values and Singular Vectors Adaptation, is a parameter-efficient fine-tuning method designed to optimize the essential singular values and vectors of large language models, significantly improving performance on various benchmarks compared to LoRA. Key features: - PiSSA achieves faster convergence and better performance than LoRA on five common benchmarks. - PiSSA reduces quantization errors in LLaMA 2-7B by 18.97% on GSM8K, leading to improved fine-tuning performance. - PiSSA is compatible with LoRA's advantages, including parameter efficiency and compatibility with quantization. Tech stack: PiSSA is implemented using Jupyter Notebook, leveraging the Hugging Face library for model fine-tuning and dataset management. Why it matters: PiSSA offers a significant improvement in performance on large language models, particularly on benchmarks like GSM8K, where it outperforms LoRA by up to 5.16% accuracy.
Python 432 stars

Rongjiehuang/ProDiff

PyTorch Implementation of ProDiff (ACM-MM'22) with a Extremely-Fast diffusion speech synthesis pipeline

### Overview: ProDiff: A Highly Efficient Text-to-Speech Model for Industrial Deployment ProDiff is a PyTorch implementation of a conditional diffusion probabilistic model designed for generating high-fidelity speech efficiently. This model, introduced at the ACM Multimedia Conference in 2022, combines a ProDiff acoustic model with a FastDiff neural vocoder to create a highly efficient text-to-speech pipeline. The model is particularly suited for industrial deployment due to its extremely fast inference times. ### Key Features: - **Extremely-Fast Inference**: ProDiff achieves near real-time processing, making it ideal for applications requiring high throughput. - **Tutorial and Code Base**: Comprehensive documentation and code examples are provided to facilitate easy integration and experimentation. - **Supported Diffusion Mechanisms**: The model supports guided diffusion, allowing for more advanced text-to-speech applications. ### Tech Stack: - **Python**: The primary programming language used for the implementation. - **PyTorch**: Framework used for the neural network architecture. - **Librosa**: Library for audio processing and analysis. - **NATSpeech**: Natural Language Toolkit for Speech Synthesis. ### Why It Matters: ProDiff offers a significant improvement in speech synthesis quality and efficiency compared to traditional methods. Its ability to generate high-fidelity speech in real-time makes it particularly suitable for industrial applications where speed and accuracy are critical. The model's design also opens up new possibilities for advanced text-to-speech applications, making it a valuable tool in the field of speech synthesis. ### Repository Details: - **Repository**: Rongjiehuang
Python 432 stars

THUDM/AlignBench

大模型多维度中文对齐评测基准 (ACL 2024)

### Overview: AlignBench 是一个全面评估中文大语言模型对齐性能的评测基准，旨在解决现有评测基准无法准确反映模型真实表现的问题。通过构建人类参与的数据构建流程和多维度、规则校准的模型评价方法，AlignBench 提供了高可靠性和可解释性的评测结果。 ### Key Features: - **全面评测**：涵盖基本任务、中文理解、综合问答、写作能力、逻辑推理、数学能力、角色扮演和专业知识等多维度。 - **多维度分析**：采用多维度、规则校准的模型评价方法，包括思维链生成和多维度分析。 - **动态更新**：通过人工检查修正答案，确保评测数据的动态更新。 ### Tech Stack: - Python ### Why it Matters: - **关键因素**：与人类意图的对齐程度已成为大语言模型实际应用的关键因素。 - **挑战**：现有的评测基准无法准确反映模型在真实场景中的表现和与人类意图的对齐程度。 - **解决方案**：构建了 AlignBench，为评估模型的对齐水平提供了全面、多维度的评测基准。 ### Repository: - GitHub: [THUDM/AlignBench](https://github.com/THUDM/AlignBench) - Stars: 432 - Description: 大模型多维度中文对齐评测基准 (ACL 2024) ### 🔥 近期更新: - 更新了 AlignBench v1.1，对涉及较强事实性