Skip to content

Latest commit

 

History

History

Folders and files

NameName
Last commit message
Last commit date

parent directory

..
 
 
 
 
 
 
 
 
 
 

README.md

Pallet Defect Detection

Pallet Defect Detection provides automated quality control with AI-driven vision systems. It enables real-time pallet condition monitoring by running inference workflows across multiple AI models. It connects multiple video streams from warehouse cameras to AI-powered pipelines, all operating efficiently on a single industrial PC. This solution enhances logistics efficiency and inventory management by detecting defects before they impact operations. It is a cloud-native application composed of microservices, using pre-trained deep learning models for video analysis. This sample application offers the following:

  • High-speed data exchange with low-latency compute.
  • AI-assisted defect detection in real-time as pallets are received at the warehouse.
  • On-premise data processing for data privacy and efficient use of bandwidth.
  • Interconnected warehouse delivery analytics for quick and informed decision making and tracking.

Get Started

How It Works

You can read the overview of the architecture and logic of the application

The components and services are as follows:

  • DL Streamer Pipeline Server is a core component of the app. It receives video feed from multiple cameras (four by default, simulated with a video recording). With pre-trained deep learning models, it performs real-time object detection, classification, and tracking. It recognizes vehicles in the parking lot and sends their 2D bounding boxes to Node-Red, through the MQTT Broker. It also adds the detected bounding boxes on top of the video input, consumed by the WebRTC Server.
  • Mosquitto MQTT Broker is a message distribution service, passing data between these sends the raw coordinates of detected vehicles to Node-RED. The feedback it receives is moved to Grafana to display.
  • WebRTC Server serves video streams processed by the pipeline for end-user visualization. It is supplemented by the Coturn signaling server and passes the feed for display in Grafana.

It also consists of the following third-party microservices:

  • Nginx is a high-performance web server and reverse proxy that provides TLS termination and unified HTTPS access.
  • MediaMTX Server is a real-time media server and media proxy that allows to publish webrtc stream.
  • Coturn Server is a media traffic NAT traversal server and a gateway.
  • Open telemetry Collector is a set of receivers, exporters, processors, connectors for Open Telemetry.
  • Prometheus is a systems and service monitoring system used for viewing Open Telemetry.
  • Postgres is object-relational database system that provides reliability and data integrity.
  • Minio is high performance object storage that is API compatible with Amazon S3 cloud storage service.

Learn More

Troubleshooting