Achieving industry-leading automated data labeling with Qwen-VLM & DeepSeek-VLM
On February 6, 2025, TIGEREYE Inc. (Headquarters: Chuo-ku, Tokyo; CEO: Gaku Kamimura) announced “TigerDataGen,” a next-generation AI data annotation system that uses the latest Vision-Language Models (VLMs). The system integrates Qwen-VLM and DeepSeek-VLM, tuned as TIGEREYE’s own models, to achieve high-precision data labeling with minimal human supervision.
Key features of TigerDataGen
High-precision annotation through multi-VLM integration
Using Qwen-VLM and DeepSeek-VLM, TigerDataGen selects the optimal VLM for each task to maximize labeling accuracy and efficiency.
AI-driven interactive annotation (Human-in-the-Loop)
The AI performs labeling automatically, and human annotators only need to do minimal checks and corrections.
Self-learning annotation pipeline
A feedback loop continuously improves labeling accuracy.
Automatic integration with TIGEREYE AI models for Object Detection & Recognition
Data labeled with TigerDataGen is directly linked to TIGEREYE’s AI models, improving learning and recognition accuracy in real time.
Scalable data pipeline
Automates data annotation for a wide range of industries, including architectural blueprints, mechanical drawings, circuit diagrams, medical images, and security footage.
Industry impact and future development
TigerDataGen makes data annotation more efficient for AI developers, enterprises, and research institutions, and is expected to be used for generating AI training data, analyzing design drawings, labeling video data, and more. TIGEREYE is currently conducting a PoC (proof of concept) to verify the accuracy of automatic annotation using datasets of architectural blueprints, mechanical designs, and security footage, and plans to roll it out as a platform for enterprises in the future.
Start by telling us about the challenges you want to solve with AI
Toward workplaces where AI is simply part of everyday work.