finds.dev← search

// the find

AlexeyAB/Yolo_mark

★ 1,841 · C++ · Unlicense · updated Dec 2020

GUI for marking bounded boxes of objects in images for training neural network Yolo v3 and v2

Yolo_mark is a desktop GUI for drawing bounding boxes on images to produce training data for YOLO v2/v3 (darknet). It is a C++/OpenCV application — compile it yourself, point it at a directory of JPEGs, draw boxes, get annotation files. For anyone who needs to hand-label images for object detection and wants something lightweight without a web browser.

The workflow is genuinely minimal: drop JPEGs in a folder, run the tool, get YOLO-format .txt files out with no intermediate conversion step. Video frame extraction is built in, which saves the usual ffmpeg scripting before labeling. The keyboard shortcuts are well-chosen — number keys for class switching, arrow keys to step through images — and faster than anything click-driven. OpenCV 2.x and 3.x compatibility means it runs on older systems without upgrading your stack.

Last commit was December 2020; the project is effectively abandoned. The entire repo is a single 1700-line main.cpp with no tests, no abstraction, and no error handling to speak of — any edge case in your image set will probably just crash it silently. Setup requires manually editing .sln files and hardcoding OpenCV paths on Windows, which is a real barrier. Modern alternatives (Label Studio, CVAT, Roboflow) support YOLO export and are actively maintained, run in a browser, and handle multi-user workflows — there is no reason to start a new project here unless you specifically need an offline C++ binary.

View on GitHub → Homepage ↗

// want more like this?

We dig through GitHub every week and send a few repos picked for what you actually care about — each with an honest take like this one.

Get finds in your inbox → Search again →