Sample frames for this project are available under NDA. Write to bilal@wortel.ai and we will share the relevant before and after pairs.
Challenges
- Annotating extremely small plates, with a median plate width of 91 px on 4,000 px images and the smallest at 12 px.
- Rotated quadrilaterals rather than axis-aligned boxes were needed so angled plates are enclosed tightly without including bumper and paintwork.
- Genuine plates had to be distinguished from phone numbers on van signage, permit stickers, house numbers and reflective badges.
- Plates that were partially occluded, clipped by the image edge, or seen at extreme angles required careful handling.
- A consistent corner ordering across thousands of polygons was needed to keep the data usable for key point training.
- Disputed annotations where the model and the original labels disagreed called for repeat review passes.
Related work
Face Blur & Picture Frame Detection
Picture frames masked and faces boxed in property photos, separating real faces from artwork for automatic privacy blurring.
Sky Segmentation & Mask Preparation for Sky Replacement
Refined binary sky masks for property photographs, holding clean edges around rooftops, trees and poles for realistic sky replacement.
Bathroom Surface Segmentation
Tiling, sanitaryware and vanity units separated per surface, through the glare and reflections bathrooms are full of.