Upload videos recorded by dash cams. The interface allows you to add video and frame specific metadata.
Deep neural networks, are used to perform object detection to recognized, cars, pedestrians and signs.
Extracted objects, along with entire frames and crops, are indexed using deep features. The feature vectors are used for retrieval by query image.
Benefit from cheap storage and data processing by using spot instances with GPUs as needed for indexing and detection.
Videos, frames, object bounding boxes and feature vectors are stored in a consistent & resuable format.
Record additional metadata such as vehicle speed, regions of interest, GPS location etc.
Use visual search along with metadata for building complex query patterns.
Support for easy integration with new deep learning models for various tasks via an extensible queue based system.