OpenThinkIMG enables modular orchestration of independent vision tools for enhanced inference workflows using PyTorch and service-based architecture. Clear quickstart included.
osx-proxmox-next automates macOS VM setup on Proxmox 9 hosts with a Python TUI, CPU auto-detection, caching, and a unique kernel patch for Apple Services compatibility.
PlatformIO Core replaces complex embedded firmware toolchains with a unified, declarative build system supporting 1,500+ boards and integrated testing workflows.
ReconViaGen reconstructs high-quality 3D objects from multiple images using a two-stage diffusion model. It runs inference on consumer GPUs and supports modular experimentation.
Roxum IDE turns Android devices into portable polyglot development environments by bundling language servers and compilers inside a Flutter-built app. It supports Rust, Go, Lua, and Dart with integrated Git and terminal workflows.
rPPG-Toolbox is a Python platform for benchmarking camera-based physiological sensing algorithms, unifying classical unsupervised and modern neural methods across multiple datasets with config-driven experimentation.
SegMAN blends State Space Models and Neighborhood Attention within a hybrid encoder-decoder for semantic segmentation, balancing long-range context with local detail. It achieves competitive mIoU on ADE20K with models from 6.4M to 92.6M parameters.
SegmentAnything3D transfers Meta’s 2D Segment Anything Model to 3D point clouds without 3D training, using depth-based mask projection and graph-based merging for zero-shot 3D segmentation.
Sketch2Motion converts static images to animated SVG-based videos using Manim with four parameterized drawing styles via a Gradio web UI. Practical animation pipeline explained.
StoryGen-Atelier automates video creation from storyboards using Google Vertex AI’s Veo model and Gemini API, combining Node.js backend and frontend with ffmpeg for media processing.
subfinder is a Go-based passive subdomain enumerator using modular sources and concurrent validation to discover valid subdomains stealthily and efficiently.
Supavec is a self-hostable RAG platform in TypeScript that cuts OpenAI embedding costs by 65% via batch processing, boosts recall by 12 points with chunk tuning, and achieves 210ms P95 latency through hybrid filtering.
SPR is an open-source router OS delivering per-device WPA3 passphrases, micro-segmented /30 subnets, and zero-trust policies in Docker containers on Linux.
Surge is a terminal download manager in Go that centralizes downloads in a daemon with multi-connection parallelism and multi-mirror failover, outperforming aria2c, curl, and wget.
TalkingHead is a JavaScript class that animates 3D avatars with real-time lip-sync and multilingual speech synthesis entirely in the browser using Three.js and WebGL.
Transcribe-Anything offers a unified CLI and Python API for Whisper AI transcription, selecting optimized backends for CUDA, Apple Silicon, or CPU. It adds unique speaker diarization and supports GPU Docker deployment.
A detailed guide to hacking HP HSTNS-PL11/PL30/PD44/PD52 1200W server PSUs for adjustable bench power. Offers hardware mods and firmware patches with deep MCU reverse engineering.
Violin is an open-source video translation pipeline integrating Whisper speech recognition, LLM translation, TTS voice synthesis, and video remuxing. It supports CLI, web app, and Claude Code skill interfaces.
Voicebun is an open-source platform for building real-time AI voice agents using Next.js, Supabase, LiveKit, and OpenAI. Deploy voice-powered AI agents from plain English descriptions.
