Enhancing surveillance AI accuracy through scalable image annotation

See how Opcito annotated 15,000 images in four weeks to train a vehicle detection model for a surveillance AI company.

Engagement Details:

Opcito partnered with a leading surveillance and physical security company that needed an AI model to detect cars, trucks, buses, and bikes for automated traffic monitoring. Opcito set up a collaborative annotation workflow on Label Studio, with a dedicated team of annotators and reviewers and batch level quality checks, and delivered a dataset ready for YOLOv5 training.

Benefits

• 15,000 images annotated in 4 weeks

• 93% mAP achieved on vehicle detection

• 70% reduction in manual effort
• No licensing cost