Camera-Based Asset Tracking Using AI and OCR - Youtube Video

Prosper YardVision AI is a next-generation, camera-based asset tracking platform that replaces RFID + GPS with computer vision, OCR, and Edge AI. It delivers real-time, precise yard visibility using standard IP cameras and the Prosper EdgeBox, with accuracy that meets or exceeds traditional RFID systems—at a lower total cost.

High-accuracy yard inventory - no RFID tags, no RFID readers, no heavy hardware

Edge-First, Cloud-Lean Architecture – All heavy video, AI inference, and OCR processing runs on the Prosper EdgeBox, with only lightweight events and metadata sent to the cloud—eliminating continuous video streaming and expensive cloud GPU costs, and delivering massive savings on infrastructure and bandwidth.

Key Solutions Powered by YardVision AI

🚛 Yard & Trailer Inventory

AI vision-based real-time visibility for trailers, containers, trucks, vessels, and yard equipment.

📹 Video AI & Vision Analytics

Real-time object detection, multi-object tracking, & visual intelligence at the edge.

📦 AI Asset Tracking Management

Live asset tracking using Camera + AI, without RFID tags or readers.

⏱️ Gate Arrival, Departure & Dwell Time Intelligence

AI vision for gates, dock doors, and vehicle movement analytics.

How It Works

Edge AI Camera Capture

Spotter-truck or fixed IP cameras stream video to the Prosper EdgeBox AI.

Smart Detection & OCR

YOLO-based AI detects assets; OCR (CRNN / PP-OCR) reads container, trailer, or unit IDs automatically.

Precise Spot-Level Location

High Accurate GPS + Homography + yard map polygons convert video frames into exact parking spot positions—no GPS drift.

Instant Event Records

Each event stores asset ID, timestamp, spot/GPS, and image snippet, locally at the edge and synced to cloud or on-prem systems.

Plug-and-Play Integrations

One-click connectors for YMS, WMS, TMS, ERP, and message buses (REST, Azure Service Bus, Kafka, MQTT).

Why YardVision AI?

💰 Edge AI = Lower Cloud Cost

🎯 Spot-level accuracy using vision, not GPS guesses

💰 Lower CapEx & OpEx vs RFID + Tags

✅ No tags. No readers. No manual scans

⚡ Real-time inventory with visual proof

🔒 Edge-first processing for privacy & reliability

Hardware Highlights

AI Module	NVIDIA Jetson™ Orin™ NX 16 GB
AI Performance	Up to ~100 TOPS (Tensor Operations per Second)
Memory	16 GB LPDDR5
Storage	128 GB NVMe SSD
Video Outputs	HDMI 2.1
Camera Inputs	2× CSI
Connectivity	4× USB 3.2, RJ45 Ethernet, CAN, GPIO
Expansion	M.2 Key E (wireless), M.2 Key M (SSD & modules)
OS	Linux + NVIDIA JetPack
Deployment	Edge / On-Premises

FEATURES SPREAD

1	Powered by NVIDIA Jetson Orin™ NX (16 GB) with up to ~100 TOPS of AI acceleration for deep learning, vision, robotics and multi-sensor processing.
2	High AI throughput with low power consumption.
3	Real-time inferencing with deterministic response.

1	4× USB 3.2 ports for peripherals and cameras.
2	HDMI 2.1 for local display output.
3	Gigabit Ethernet (RJ45) for networking.
4	M.2 Key E / Key M slots — connect Wi-Fi / LTE / 5G, SSD, and expansion modules.
5	CAN, GPIO, dual CSI camera interfaces.

1	Pre-installed NVIDIA JetPack (Linux + Jetson SDK)
2	Support for CUDA, TensorRT, DeepStream
3	ROS / ROS2 frameworks for robotics
4	Python & C++ ready for custom models