Изображения
Interactive-Image-Background-Remover
Point-and-click AI background removal with Segment Anything and BiRefNet
★ 59 звёздCustom
Смотреть руководство →SAM 2 — проект с открытым исходным кодом из категории «Изображения» от facebookresearch (facebookresearch/sam2 на GitHub). У него 20 061 звезда на GitHub и 2 581 форк, он распространяется по лицензии Apache-2.0 и последний раз обновлялся на GitHub 30 мая 2026.
SAM 2 — Meta's Segment Anything Model 2: promptable segmentation for images and video, Apache-2.0
SAM 2 needs Python 3.10+, torch 2.5.1+ and torchvision 0.20.1+ — install those from the PyTorch site before anything else.
Clone the repo and install it in editable mode; add the notebooks extra for the example notebooks.
Download a SAM 2 checkpoint, then run image or video prediction with the predictor from the example notebooks.
Point, box or mask prompts select exactly the object you mean — no training on your subject required.
The step beyond SAM 1: the same model tracks a segmented object across video frames.
Image prediction and video prediction notebooks show the full workflow, with jupyter and matplotlib as the only extras.
Segmentation masks feed background removal, object cut-out, inpainting and video effects pipelines.
The README install route: clone, then editable install.
git clone https://github.com/facebookresearch/sam2.git && cd sam2
pip install -e .Required for the example notebooks (jupyter and matplotlib).
pip install -e ".[notebooks]"Checkpoints are separate from the code — download them per the README Getting Started section, then run image or video prediction.
https://github.com/facebookresearch/sam2#getting-startedMeta's Facebook Research team, as the successor to the Segment Anything Model.
Yes — Apache-2.0 licensed code; checkpoints carry their own licence terms.
SAM 2 adds video: it segments and tracks objects across frames, where SAM 1 handled single images.
For practical use, yes — the install compiles a custom CUDA kernel, and the README recommends a GPU machine (WSL on Windows).
Install PyTorch 2.5.1+ and TorchVision, clone and pip install -e ., download a checkpoint, and follow the image or video prediction notebooks.
Point-and-click AI background removal with Segment Anything and BiRefNet
Real-ESRGAN — practical blind super-resolution for real-world photos and video: restores and upscales degraded images with models trained on pure synthetic data, BSD-3-Clause
Background Remover lets you Remove Background from images and video using AI with a simple command line interface that is free and open source.
Enhance images in JavaScript — upscale, denoise, deblur and dehaze with AI
Open-source skill that turns coding agents into AI image and video creators
Browser-only AI interface for UI mockups, covers, logos and social images
Другие проекты из категории «Изображения» с наибольшим числом звёзд в каталоге LibreAI — реалистичный список альтернатив, если SAM 2 вам не подходит.
Stable Diffusion web UI
The node-graph powerhouse for precise, professional AI image and video generation.
High performance self-hosted photo and video management solution
PhotoCraft — a clean-room Photoshop reimplementation in pure Rust: layers, masks, adjustment layers, type, vectors, brushes and real PSD files, offline, MIT / Apache-2.0
Focus on prompting and generating
Соседние проекты из категории «Изображения» в каталоге, ближайшие по числу звёзд к SAM 2 — продолжайте изучать полку по обе стороны от него.
How Python does AI. Agents, realtime voice, image generation, embeddings. Every model, every interface, typed end to end.
[NeurIPS 2022] Towards Robust Blind Face Restoration with Codebook Lookup Transformer
Rembg is a tool to remove images background
Stable Diffusion with Core ML on Apple Silicon
The classic PyTorch implementation of CycleGAN and pix2pix for image-to-image translation
DeepSeek's Janus series - unified multimodal models that both understand and generate images
Бесплатно и открыто — забирайте исходный код, ставьте звезду репозиторию и начинайте создавать.