ImgCompress

ImgCompress is a self-hosted image processing toolbox built by Karim Zouine and distributed as a single Docker image under the GPL-3.0 license. It handles format conversion across 70+ input types, bulk compression, AI-powered background removal, and PDF export — all running locally on your own hardware with no external API calls or cloud dependency. The project is written primarily in TypeScript and has been accepted into the Awesome Self-Hosted curated list and listed as an official Coolify service, according to the project homepage.

What It Is

ImgCompress is a self-hosted, containerized image processing tool that replaces a collection of single-purpose converters and cloud-based tools with one Docker image. It targets self-hosters, homelab operators, and privacy-conscious users who want to process images — including unusual formats like HEIC, PSD, TIFF, EPS, and RAW — without uploading files to third-party servers. The web UI runs at localhost:3001 after a single docker run command, and a full CLI is also bundled in the same image for scripting pipelines.

Core Capabilities

• 70+ format support: 10 formats are fully tested and confirmed (including HEIC, HEIF, PNG, JPG, ICO, EPS, PSD, PDF, AVIF, and WebP); 60+ additional formats are supported and likely to work.
• Local AI background removal: A bundled AI model runs entirely on the host CPU — no API key, no subscription, no network call required. The model is included in the image so no separate download is needed.
• Batch compression: Multi-core parallel processing lets users shrink entire photo libraries in one job with control over quality targets and output format.
• Smart PDF creation: Converts image batches to structured A4 PDFs with automatic pagination, handling long screenshots cleanly across pages.
• Air-gap ready: Once the image is pulled, it runs with no internet connection. No license checks, no telemetry, no expiry.

Privacy and Deployment Model

The tool is designed around a "privacy by design" principle: conversions, compression, and AI inference all run locally, files never leave the host machine, and the container produces no outbound network traffic after startup. There is no analytics, crash reporting, or feature-flag system phoning home. This makes it suitable for private networks, NAS drives, and homelabs where data residency matters. The entire stack is open source under GPL-3.0, so it is free to use, modify, and distribute.

Setup Path

Deployment takes roughly 60 seconds: pull the image with docker run -d --name imgcompress -p 3001:5000 karimz1/imgcompress:latest, then open localhost:3001 in a browser. ImgCompress is also available as an official service in the Coolify self-hosted deployment platform, allowing one-click addition from the Coolify dashboard.

Update: v0.6.1

The latest release is v0.6.1, published on 2026-04-17. The repository was last pushed on 2026-05-17 and last updated on 2026-05-20, indicating active maintenance. The project was created in December 2024 and has accumulated 214 stars and 22 forks on GitHub as of the latest data.