Maschinelles Lernen · GitHub

[CVPR 2026] Multinex: Lightweight Low-light Image Enhancement via Multi-prior Retinex

A minimalist SOTA LaTeX OCR model with only 20M parameters, running in browser. Full training pipeline available for self-reproduction. | 超轻量SOTA LaTeX公式识别模型，仅20M参数量，可在浏览器中运行。训练全流程代码开源，以便自学复现。

• Free and open-source.
• Fast and lightweight inference.

This is an open source project (formerly named Listen, Attend and Spell - PyTorch Implementation) for end-to-end ASR implemented with Pytorch, the well known deep learning toolkit.

• On-the-fly feature extraction using torchaudio as backend
• Character/subword2/word encoding of text

An optimized handwritten digit recognition model delivering exceptional accuracy, high-speed performance, and compatibility with low-end hardware, supported by comprehensive documentation, experiment tracking, data-driven decision-making, effective troubleshooting, and a modular project structure.

• Maximum execution speed
• Highly robust optimization

Handbuch für Deep-Learning-Interviews zu Mathematik, maschinellem Lernen, Deep Learning, Computer Vision, natürlicher Sprachverarbeitung, SLAM und weiteren Bereichen.

Interview = 简历指南 + 算法题 + 八股文 + 源码分析

Fast, offline OCR for Node.js & C++. PP-OCRv6 with Core ML / WebGPU hardware acceleration — recognize text in images with confidence scores & coordinates. npm: @arcships/light-ocr

Monocular, One-stage, Regression of Multiple 3D People and their 3D positions & trajectories in camera & global coordinates. ROMP[ICCV21], BEV[CVPR22], TRACE[CVPR2023]

Tools to Design or Visualize Architecture of Neural Network

Fast Volleyball Tracking Inference: Real-time volleyball ball detection and tracking at 100 FPS on CPU (Intel i5-10400F). Powered by an optimized ONNX model, outputs ball coordinates to CSV, with optional video visualization. Ideal for sports analytics and computer vision research.