Case study — Timber & Forestry
TimberVision: CNN-based timber classification
An Australia-based client seeking an AI-powered system to classify timber species using image recognition, with real-time predictions through a web interface and mobile app.
Industry
Timber & Forestry
Location
Australia
Engagement
AI/ML product build
What we built
Built an end-to-end AI-powered timber image classification system using deep learning (a ResNet50 CNN) to identify timber species, grain patterns and quality grades from photographs.
Read more
Delivered a scalable backend API in FastAPI for model inference, a web interface in React, and a cross-platform mobile app in React Native for real-time, on-site timber inspection.
Combined transfer learning, optimised deployment and focused UI design to make timber analysis faster, more consistent and less dependent on manual expert inspection.
Challenges
- No publicly available image dataset for timber species existed to train on
- Prediction accuracy dropped between clean training images and real field photographs
- The backend needed to handle simultaneous image uploads at scale
- The mobile app needed to remain usable with limited or no connectivity on-site
Technologies
- ResNet50 CNN
- FastAPI
- React
- React Native
- PyTorch
- AWS
The result
Delivered on time, on scope, on budget.
- Timber specification became an automated step rather than a manual one
- Classification accuracy and consistency improved over manual inspection
- Reduced reliance on expert inspectors for routine classification, lowering cost
- Real-time, scalable analysis available directly from a phone on-site
Have a similar problem?
Tell us what you're trying to solve. We'll tell you plainly whether it's a good fit and what it would take.
Or reach us directly at info@ontogenesistechnologies.com · +91-9999101853