Cross-platform desktop app for AI-assisted image, video, and point cloud annotation.
Label text, images, video, point clouds, and multimodal data with deep learning models.
X-AnyLabeling is a lightweight desktop application designed for annotating multi-modal data using state-of-the-art deep learning models. It suits engineers and researchers needing automated labeling across text, images, video, and 3D point clouds.
The application combines traditional shapes like polygons and bounding boxes with modern vision foundation models, OCR engines, and segmentation tools. It executes local or remote inference via ONNX Runtime, TensorRT, or vLLM to automate dataset labeling workflows.
It suits computer vision engineers and data labelers preparing datasets for object detection, segmentation, and document parsing. It does not suit teams looking for a purely cloud-native web application without a desktop footprint.
Annotate text, images, video, 3D point clouds, and multimodal data inside a single desktop interface.
Integrate state-of-the-art models like SAM, YOLO, Grounding DINO, and PP-OCR for automated inference.
Import and export data across standard formats including COCO, VOC, YOLO, DOTA, and MOT.
Perform 3D point cloud detection and segmentation natively within the cross-platform application.
Connect custom models and heavy compute resources using the accompanying X-AnyLabeling-Server.
The software is free under its open-source license, with no paid hosted options provided by the project. The only cost is running it yourself on local hardware or provisioning your own remote inference servers.
873 open remote AI-training roles, updated November 15, 2026; most pay $20–$170 an hour. Zekai lists them free; the application goes to the employer’s platform.
You can import and export standard formats including COCO, VOC, YOLO, DOTA, MOT, MASK, PPOCR, and ShareGPT.
Yes, the tool supports custom model integration and secondary development through its documentation and CLI.
No, it operates locally on your desktop, though running certain foundation models locally demands adequate GPU hardware.
See Zekai first in Google