Digital AIUpdated 24 October 20268 min read

TensorFlow Lite in Manufacturing: Edge Models Guide

TensorFlow Lite, now called LiteRT, lets engineers shrink a trained model and run it on an edge device beside the machine. This guide covers the workflow from data to deployment, quantisation, TensorFlow versus PyTorch, and the monitoring that keeps a model honest.

TensorFlowLiteRTDeep LearningEdge DeploymentMachine Learning
Hands pointing at a process-analysis workflow on a laptop screen beside a test rig

Key takeaways

  • TensorFlow Lite has developed into LiteRT, a runtime for edge devices that also accepts models exported from PyTorch and JAX in the .tflite format.
  • A dependable pipeline runs from a defined decision through honest data splits, conversion and quantisation to monitoring and retraining.
  • Post-training quantisation can make a model about 4 times smaller, but accuracy can fall, so re-test escape and false reject rates.
  • Start with a simple baseline; many sensor problems do not need deep learning.
  • Treat a deployed model as controlled equipment: version it, test it on a reference set, log results and define a fallback.

TensorFlow Lite is Google's toolkit for running trained machine learning models on small devices, and it has since developed into LiteRT, which Google describes as a high-performance, cross-platform inference runtime for edge devices. In manufacturing, you train a model with TensorFlow or another framework, convert it to the compact .tflite format, and run it on an industrial PC, gateway or microcontroller next to the machine.

What is the workflow from data to a deployed factory model?

Deploying deep learning is mostly engineering around the model. A dependable pipeline has seven steps:

  1. Define the decision. "Flag a bearing likely to fail within a week" or "reject cracked castings." A vague goal gives a vague model.
  2. Collect and label data. Record sensor signals or images with the operating context and the engineers' judgement of what was good or bad.
  3. Split data honestly. Hold out whole days, shifts or machines for testing, not random rows, otherwise the model looks better than it is.
  4. Train and evaluate. Build a baseline first, such as a simple threshold or a gradient-boosted model, so you know whether deep learning adds anything.
  5. Convert and optimise. Export to .tflite and reduce size and latency with quantisation.
  6. Deploy and integrate. Run on the edge device, expose results over MQTT or OPC UA, and let the PLC or SCADA act on them.
  7. Monitor and retrain. Track input data, predictions and outcomes; retrain when the process or the camera changes.
Simplified diagram of a neural network learning to tell two object types apart
Simplified diagram of a neural network learning to tell two object types apartImage: Mikael Häggström, CC0, via Wikimedia Commons
Introducing TensorFlow Lite (Coding TensorFlow)

What does quantisation do, and what are the risks?

Quantisation stores a model's numbers at lower precision so it is smaller and faster. Google's documentation describes three post-training options:

  • Dynamic range quantisation: weights become 8-bit, about 4 times smaller with roughly 2 to 3 times faster CPU inference, and no calibration data is needed.
  • Full integer quantisation: weights and activations become integers, about 4 times smaller with larger speed-ups and support for accelerators and microcontrollers. It needs a representative dataset of roughly 100 to 500 samples.
  • Float16 quantisation: about half the size, mainly useful for GPU acceleration.

The same documentation warns that accuracy can fall, particularly for smaller networks, so you must test the quantised model against your acceptance limits, and use quantisation-aware training if the drop is too large. For an inspection model, re-check the escape rate and false reject rate after conversion, not just overall accuracy.

Should you use TensorFlow or PyTorch?

Either will do. TensorFlow with Keras has mature deployment tooling and a long history in edge deployment. PyTorch is widely used in research and many open-source vision models. LiteRT supports direct export and quantisation from PyTorch, TensorFlow and JAX into the .tflite format, so the framework you train in no longer locks you to one runtime. Pick the one your team and your data science tools already know, and focus effort on data and validation instead.

Also consider whether deep learning is needed at all. For many sensor problems, such as vibration features on a motor, a simpler model is easier to explain, runs on a tiny device and is quicker to validate. Our predictive maintenance guide shows how far simple methods go.

What manufacturing problems suit TensorFlow-style models?

  • Image classification and defect detection for surface flaws, assembly checks and sorting.
  • Time-series classification of vibration, current or acoustic data for condition monitoring.
  • Anomaly detection using autoencoders trained on normal operation.
  • Tiny models on microcontrollers that listen for a bearing noise or detect an unusual vibration signature without a network connection.

What should you check when you convert a model?

Conversion is where many projects lose time. Confirm that every layer in your model is supported by the runtime on your target device, since unusual custom operations may not convert. Compare outputs of the original and converted models on the same few hundred inputs and look at the largest differences, not just the average. Measure latency on the real device, not on your laptop, with the camera or sensor feed attached and other applications running. Check memory use over hours, not minutes, to catch leaks. If the target is a microcontroller, check that the model fits in flash and RAM with room to spare for the rest of the firmware.

Keep the preprocessing identical between training and deployment. A different image resize method, colour order or sensor scaling is a classic source of models that work in the notebook and fail on the line.

How do you keep a deployed model trustworthy?

Treat it as a controlled piece of equipment. Version every model and record the training data and settings that produced it. Test it on a fixed reference set before each release. Log inputs, outputs and the human decision that followed so you can measure real-world performance. Alert when the input data drifts from what the model saw in training, such as a new product, a replaced camera or a sensor recalibration. Define a fallback: if the model is unavailable, the process should continue safely on conventional control or manual inspection.

Connecting a model to control equipment should follow the site's management of change process. Our edge computing guide covers the hardware and security side.

When is a gateway enough, and when do you need a GPU?

A model that classifies a few vibration features once a second runs happily on a modest industrial PC or even a microcontroller. An image model analysing several frames a second at full resolution will usually want a GPU or an accelerator. Work out the numbers first: images per second, image size and acceptable latency, then test a candidate device with your actual model. Buying hardware before measuring is the most common and most expensive mistake in edge AI projects. NVIDIA's Jetson range, for example, spans entry-level modules to high-end ones, and its own figures are not comparable across generations, so benchmark with your workload.

What skills and jobs does this lead to?

ITJobsWatch reported a median UK salary of £75,000 for permanent machine learning roles across 2,526 advertisements in the six months to 23 October 2026. Most of those roles are in software and data teams, but industrial employers increasingly need engineers who can take a model from a notebook to a machine. The scarce skill is deployment and monitoring, not building yet another classifier.

How to learn TensorFlow and edge deployment for free

Start free

Begin with the free Deep Learning with TensorFlow course, try the first five free days of the Machine Learning Engineer job simulation, and look at the AI and machine learning jobs page. If you want a self-paced route with five graded industrial projects, see the EDWartens AI Module for its current price and syllabus.

Start free

Learn AI and machine learning free, then practise the job

Free job simulation

Machine Learning Engineer

Train, evaluate and document a model, and catch target leakage.

The first five days are free, no card needed.

Live UK jobs

AI and ML jobs in the UK

Updated daily, with licensed visa sponsors marked.

Browse all engineering jobs

Frequently asked questions

What is TensorFlow Lite used for?

It is used to run trained models on small devices such as phones, gateways and microcontrollers. In factories that means running inspection or condition-monitoring models on an edge device beside the machine. It is now part of Google's LiteRT runtime.

Is TensorFlow Lite the same as LiteRT?

LiteRT is built on TensorFlow Lite and is the name Google now uses for its on-device inference runtime. It supports models exported from TensorFlow, PyTorch and JAX in the .tflite format, and Google provides a migration guide.

Should I learn TensorFlow or PyTorch for industrial AI?

Either works. Choose the framework your team uses and concentrate on data quality, evaluation and deployment. Because LiteRT accepts models from several frameworks, you are not locked in.

Does quantisation reduce accuracy?

It can, especially in small networks. Google's documentation says post-training quantisation can make a model about 4 times smaller with little accuracy loss, but you must test the converted model against your acceptance limits and consider quantisation-aware training if the drop is too large.

How do I keep a factory model from going stale?

Monitor input data and predictions, compare with human decisions, re-validate after changes in process, product or camera, and retrain on a schedule. Assign an owner for the model after launch.

Sources

  1. LiteRT overview, Google AI Edge (formerly TensorFlow Lite)
  2. Post-training quantization, LiteRT documentation
  3. NVIDIA Jetson modules
  4. Machine learning jobs in the UK, ITJobsWatch (checked 24 October 2026)

Share this article

About the author

EDWartens Team

EDWartens Instructional Team

The EDWartens instructional team brings together UK-based automation engineers, certified Siemens and Rockwell specialists, and industry practitioners with decades of combined experience in PLC programming, SCADA engineering, and industrial AI.

View all articles by EDWartens

Ready to become job-ready?

Start with a free course today, or train live in Siemens PLC, HMI and SCADA on the Professional Module.

Related articles

Digital AI

The Future of SCADA in Industry 4.0: What Is Changing

SCADA is moving to web clients, open protocols such as OPC UA and MQTT, redundant servers with central archives, version-controlled engineering and tighter network security. UK permanent SCADA adverts rose to 632 in the six months to 21 October 2026, from 242 a year earlier, with a median salary of £55,000 (ITJobsWatch).

Read article
Digital AI

Predictive Maintenance Explained: A Practical Guide for UK Plants

Predictive maintenance schedules repairs from the measured condition of a machine rather than the calendar. Siemens' 2024 survey found nearly half of large industrial firms now have dedicated predictive-maintenance teams, twice the share of 2019. This guide compares the four maintenance strategies, the sensors involved, where PLCs and SCADA fit, and a realistic adoption path.

Read article
Digital AI

Machine Learning for Quality Control in Manufacturing

Machine learning for quality control uses trained models to classify defects, flag anomalies and predict out-of-spec parts from images and process data. This guide covers the three approaches, how to measure them honestly, and a UK worked example.

Read article
Digital AI

Machine Vision Inspection: Computer Vision Guide

Machine vision inspection uses cameras, lighting and software to check parts automatically. This guide explains how classic and deep-learning inspection differ, what a system contains, how it talks to the PLC, and how to learn the skills in the UK.

Read article