Industrial & Inspection

Edge AI Food Quality Control for Fast-Food Kitchens

Real-time pass/reject checks as each dish is plated, made on the edge device with no kitchen video sent to the cloud.

99.2%

Dish recognition
accuracy

79–99%

Per-ingredient
counting accuracy

6 weeks

From brief to a working
on-device pipeline

Customer Story

A multi-site food-service operator is standardising dish quality, where every plate has to meet the same spec regardless of location or cook. Blackthorn AI built an edge AI system for food quality control: computer vision counts each component and a rule engine makes the pass/reject call.

Blackthorn AI validated the pipeline on the target NXP edge device for real-time operation and handed it over with the labelled dataset and deployment documentation, six weeks after the brief.

Computer Vision, Edge AI, MLOps, Data Annotation

Services

Netherlands

Location

Edge computer vision · 6 weeks

Engagement

Challenge

A pass-or-reject quality check at the moment of plating, on-device, not in the cloud

Quality control in a fast-food kitchen is a real-time decision: each plate is served or remade within seconds, and a wrong call means food waste, a slower ticket or a complaint. The client needed to automate that check across sites without streaming video to the cloud: continuous upload is impractical at scale, and cloud round-trips are too slow and too costly. The rules are strict and countable: 14 sausages spread evenly to the edge of a pizza, two diagonally cut toast slices, an intact egg yolk, 85 g of beans. The system has to count components, judge portions, apply spatial rules and decide.

Solution

A two-stage edge AI pipeline for real-time food quality control

Blackthorn AI built the check to run in real time, entirely on the edge device. In the first stage, quantised detection and segmentation models, fine-tuned on the client’s dishes and running on the NPU, detect and count every component: bacon, sausages, hash browns, toast and beans. In the second, a rule engine on the same device checks that output against the kitchen’s QA rules and returns pass or fail with the reason attached.

Camera mounted under a shelf above the stainless-steel pass, wired to the edge AI device, with an empty plate on the counter
Camera and edge device at the pass
Green detection boxes around the egg, bacon, sausage, beans, hash browns and toast on a full breakfast plate
Component detection on a breakfast plate
Green detection boxes around the pepperoni slices on a pizza, the input for the topping-count rule
Pepperoni detection for the topping count
Tech Stack

Built with Python, PyTorch and C++: YOLO-World detection and DDRNet segmentation on an NXP i.MX NPU, plus a custom rule engine.

Python
PyTorch
C++
YOLO-World
DDRNet
NXP i.MX NPU
Roadmap

Project duration

Six phases over six weeks, from QA rules to handover.

Phase 1 QA rule definition

Completed

Translated the client’s plating standards into explicit, countable QA rules per dish and component (pizza, breakfast, bacon, toast, eggs).

Phase 2 Dataset construction

Completed

Assembled about 500 images from client and open sources, covering every component and every defect class.

Phase 3 Annotation & balancing

Completed

Labelled images manually and semi-automatically against visual evidence, balanced across dishes and good/bad cases so the models learned the rules, not the common cases.

Phase 4 Model training & optimisation

Completed

Trained and fine-tuned the detection, counting and segmentation models on the client’s dishes and locked in the deployed configuration.

Phase 5 Edge pipeline & deployment

Completed

Built the two-stage pipeline, with computer vision on the NPU and the rule engine alongside it, and validated it on the NXP edge device for real-time operation.

Phase 6 Handover

Completed

Delivered the working pipeline, the labelled dataset and deployment documentation.

Team Size

8 Qualified Specialists
1 × Computer Vision Technical Lead
2 × Computer Vision Engineers
4 × Annotation Specialists
1 × Project Manager

What the Pipeline Delivered

The whole decision runs on the edge device: computer vision detects and counts every component on the NPU, and a lightweight rule engine on the same device applies the kitchen’s QA rules, with no cloud round-trip or video upload.

99.2%

Dish recognition accuracy

The on-device computer vision stage identifies the dish with 99.2% accuracy, giving the rule engine a reliable starting point for every plate.

79–99%

Ingredient counting accuracy

Bacon, sausages, hash browns, toast and beans are counted at 79–99% accuracy depending on the ingredient, turning a plate photo into rule-checkable data.

6 weeks

From brief to working pipeline

An eight-person team delivered QA rules, a purpose-built dataset, trained models and edge deployment on the target NXP hardware in six weeks.

FAQ

Edge AI food quality control: common questions

Blackthorn AI’s system checks each plate in two stages, both on the edge device. A computer vision model on the NPU detects and counts every component on the plate. A deterministic rule engine then checks that output (what is on the plate, how many and where) against the kitchen’s QA rules, such as 14 sausages on a pizza or two toast slices. It returns pass or fail and names any rule that failed.

No. The QA rules are countable and deterministic, so a rule engine makes the decision. That is deliberate: edge NPUs are built for small convolutional networks rather than large multimodal models, so an LLM could not hit real-time latency on the device, and the explicit logic does not need one.

Yes. In this system the whole decision is made on the edge device: computer vision runs on the NPU and the rule engine runs alongside it. Neither the video nor the detection data needs to leave the device, so the pass/reject call lands in real time with no cloud round-trip. This also removes bandwidth cost and privacy exposure.

Blackthorn AI’s on-device pipeline reaches 99.2% dish recognition and 79–99% counting accuracy, depending on the ingredient. Detection and counting are largely solved. Portion estimation, such as judging 85 g of beans from a single image, is the hardest rule and the main limit on full pass/fail accuracy.

This project took six weeks from brief to a working on-device pipeline, delivered by an eight-person team. The work covered QA rule definition, a dataset of about 500 images, annotation and balancing, model training, deployment on the NXP edge device and handover. For a production-line inspection project, see our industrial computer vision quality control case study.

Discover More Related Customer Stories