When a tyre is at the end of its life, it is sent to a handling center. Depending on the characteristics of the tyre, it may or may not be retreaded. The procedures are very precise and the diagnosis must be perfectly objective.
How can AI help the diagnosis? How to speed up feedback?
Neovision first conducted a first exploratory image recognition study on a small amount of data to assess the level of achievable performance.
Second, a greater amount of data was explored, in a larger project and integrating classification challenges.
These exploratory studies have validated many benefits for our client and demonstrate the potential of automated feedback on a large number of products. Indeed, operators will be able to identify more quickly the characteristics observed in the field. This work will increase knowledge about tire use in the future.
The first study consisted of a statistical and visual study on a small volume of data. The objective was to be able to analyze the distribution of data in the different categories, in order to measure the level of achievable performance on classification issues. For this, we used our data visualization tool Tadaviz. It allowed us to predict the level of performance achievable to properly classify the characteristics present on the tires.
Once the performance level was validated, we increased the volume of data to train deep learning algorithms to achieve this performance. We worked with business experts to clarify certain characteristics and classification criteria.
Deep Learning, Computer Vision, Image Classification
« Very quickly, our teams forged bonds of trust and mutual respect that are strengthened over time. This is very noticeable in the regular exchanges that bring us together. It must be said that we have great professional successes to share, this makes us proud on both sides and makes our teams stronger.
Neovision has demonstrated active listening and a great ability to understand our expectations, our constraints and our needs; The proposed solutions are the proof, each time delivered on time and meeting our objectives. Neovision has also been a driving force thanks to the innovative solutions we adopt : Tadaviz.
Congratulations and thank you. »
Alain BOITEAU, Data Scientist,
Product Quality and Market Performance Surveillance
27 September 2023
Applications Web, Cloud, Computer Vision, Deep Learning, Industrie