Protecting Personal Data in Grab's Imagery

Summary

Grab's KartaView platform collects geotagged street imagery across over 100 countries, requiring automated obfuscation of faces and licence plates to protect personal privacy. Because off-the-shelf solutions struggled with diverse global environments and equirectangular 360-degree camera formats, Grab built a custom machine learning pipeline. The system projects varied image formats into standardized planar views, applies a YOLOv4 object detection model to locate target regions, and transforms bounding coordinates back to the original imagery for blurring. Training the detector required iterative dataset updates to accommodate edge cases like face masks and mirror reflections, paired with offline view splitting and oversampling of scarce large bounding boxes. Assessments confirmed that obfuscating these regions had minimal negative impact on downstream map feature extraction services.

Takeaways

1. 1

   To handle distortions across normal, wide-angle, and 360-degree equirectangular formats, the pipeline normalizes inputs into overlapping 75-by-75-degree planar image views before detection.

2. 2

   Dataset splitting must enforce that continuous trip images stay within the same subset to prevent geographically correlated samples from leaking between training and evaluation splits.

3. 3

   Oversampling scarce, large bounding box labels was necessary to generate proper anchor boxes and improve recall for close-up detections in anchor-based object detectors like YOLOv4.