Powered by OpenFilter, A Universal Abstraction for Computer Vision Workloads
Demo Application: Overlays subject data from DriveID, Clipper, and TextScan on processed license plate frames

Generic vision filter using Hugging Face Transformers for object detection and image classification across video streams

Real-time object detection using RT-DETR v1 architecture
A production-ready OpenFilter component that collects events from filter pipelines and reliably delivers them to a CloudEvents ingestion compatible endpoint.

Real-time object detection using RF-DETR architecture

Real-time object detection using RT-DETRv2 architecture
crops any region of interest (polygon) in a video and creates a new video from it
Exports data to MQTT, bundled in with OpenFilter
Detect motion and structural changes, and save video frames that meet specific criteria
Write a single segmented video file or rtsp stream
Show incoming topic image stream on a web server
Read a single video file, network stream or webcam until the end
Read a images from a folder or GCS bucket or S3 bucket
Record frame data to files as JSON or CSV. If the output filename ends with '.csv' the output will be CSV, otherwise JSON