CASE STUDY
Digital Watches
A globally-renowned electronics company sought our expertise to elevate their defect detection capabilities within their digital watches, analyzing for defects on the watch faces.
Mindtrace was selected to create an industry-leading AI model that seamlessly integrates into their existing model backbone, reducing disruption while revolutionizing their manual quality assessment procedures.
Challenges
- Under 300 available training data images
- Unlabelled training data
- Images contain high levels of noise
Step 2 - Model Testing
After the training phase was complete, the model was tested on further data samples, achieving a 97% defect detection rate.
Step 1 - Model Training
During the training phase, Mindtrace's Brain-Sense™ Platform successfully processed 300 training images utilizing few-shot learning due to the limited information present in the dataset.
Step 2 - Model Testing
After the training phase was complete, the model was tested on further data samples, achieving a 97% defect detection rate.
Step 1 - Model Training
Outcome
Despite the constraints of a limited dataset and the presence of noise, Mindtrace’s Brain-Sense™ platform effectively developed a bespoke AI solution that achieved a 97% defect detection rate within the training data.
Furthermore, this solution seamlessly integrates into their existing model backbone, with deployment options available for on-premises, cloud, or edge devices, ensuring minimal disruption.
AI Brains Used
1
Detection
Identifies defect types, severity & location.
1
Classification
Object identification & classification.
1
Analytics
Generates insights & recommendations.
1 Learning Samples 1% Defects Detected 1 Delivery Time