Automated Fruit Classification with Neousys Nuvo Edge Computers

Neteon

November 19, 2025

Case Studies

Overview

Modern fruit exporters need consistent, high-speed quality control—especially when microscopic damage can trigger quarantine rejections. To tackle inspection inaccuracy, labor cost, and throughput limits, an automation provider built an automated fruit classification solution powered by Neousys embedded computers and machine learning.

The system captures images of each fruit, inspects them with AI algorithms, rejects defective pieces, and distributes graded fruit by size, color, and quality on the conveyor line.

Tiny stings or pin-sized holes from pests such as codling moth are almost invisible to the human eye, but they are enough to ruin a shipment. By bringing AI inference right onto the packing line, exporters gain repeatable grading quality at industrial speed.

Key Challenges for Automated Fruit Classification

‍

| Challenge | Why It Matters | What the Edge System Must Provide | | ------------------------------------ | ---------------------------------------------------------------------------------------------------- | -------------------------------------------------------------------------------------- | | True wide-temperature operation | Previous PCs overheated in hot packing houses, throttling ANN performance and slowing sorting speed. | Rugged, fanless thermal design that sustains full load across wide temperature ranges. | | Multi-angle camera coverage | Defects may hide around the calyx and stem bowl, requiring multiple viewpoints per fruit. | Sufficient PoE/USB ports and I/O to power and synchronize several cameras and sensors. | | High compute density in tight spaces | ANN, image pre-processing, and control logic must run near the conveyor in a compact cabinet. | High-core CPU (and optional GPU) performance in a small, reliable embedded chassis. |

‍

Solution: Edge AI Machine Vision on Neousys Nuvo

The provider designed a machine vision system built around a Neousys Nuvo embedded system:

CCD cameras capture fruit from multiple angles as it passes on the conveyor.
Images are delivered via frame grabbers to the Nuvo platform.
AI models (Artificial Neural Networks, centroid and box-counting methods) evaluate fruit surface, color, and shape.
The embedded computer sends commands to the servo control system to route each item to the correct lane or reject line.

Neousys’ fanless thermal design, as used in its Nuvo fanless embedded computers, prevents the overheating and throttling the customer experienced with their previous vendor while keeping the footprint compact for line-side installation.

Advantages of Neousys Nuvo Embedded System

Compact, rugged design with rich I/O
Multiple PoE/USB ports support multi-angle vision; COM/DIO help integrate with robotic arms, gates, or AMRs for downstream material handling.
High performance Intel® Core™ processing
Newer platforms like the Nuvo-11000 Intel® Core™ Ultra 200 fanless embedded computer deliver up to 24 cores for ANN inference, image processing, and deterministic control on a single box.
Reliable 24/7 edge operation
Fanless construction and wide-temperature design allow continuous use in hot, dusty packing houses without constant maintenance.

‍