Face Blur & Picture Frame Detection

Project overview

Property and real estate images were annotated for a computer vision dataset designed to detect picture frames and human faces for privacy protection. The work involved creating accurate polygon and bounding box annotations to generate high-quality training data for an AI model capable of automatically detecting and blurring faces before property images are published.

Work performed

• Reviewed and organised the property image dataset before annotation.
• Created accurate polygon annotations for all picture frames.
• Visible human faces were annotated using bounding boxes.
• Picture frames were classified according to whether they contained visible human faces.
• Marked faces requiring automatic privacy blurring using the Face_bbox_blur class.
• Annotation consistency was maintained across all images.
• Performed quality assurance checks and corrected annotation errors before delivery.
• Delivered completed annotations according to client requirements.

What made it hard

• Detecting very small or partially visible faces inside picture frames.
• Handling reflections, shadows, glare, and low-resolution images.
• Real human faces had to be distinguished from paintings, artwork, and decorative objects.
• Creating accurate polygon boundaries around irregular picture frames.
• Consistency between frame and face annotations had to be maintained across the dataset.

Result

A high-quality annotated dataset was delivered for privacy-aware computer vision applications. The dataset was prepared for training an AI model capable of detecting picture frames, locating human faces, and automatically blurring sensitive facial information in property and real estate images.

Final deliverable

Property image annotations were completed across four annotation classes, namely Face_bbox_blur, Face_bbox, Frame_mask_without_faces, and Frame_mask_with_faces, with high annotation accuracy, consistency, and quality maintained throughout. The final dataset was delivered according to client requirements and prepared for training an AI-powered automatic face blurring system.