A semiconductor giant offering a wide range of application processors and MCUs with secure connectivity for industries like automotive, industrial & IoT, mobile, and communication infrastructure markets.
As a successful semiconductor company, client launches multiple processors and MCUs for various industrial application scenarios from time to time. With the launch of one of its processor i.MX RT series, client wanted to introduce machine learning based handwritten digit recognition application on edge into the market. For the same, client was looking for a Machine Learning expert who can help them achieve their launch goal.
VOLANSYS is working as client’s Machine Learning partner by providing deep learning expertise on their edge platform. We helped them develop a solution based on deep neural network to offer applications like digits recognition on number plates, handwritten numeric entries on banking cheques or any forms reducing the manual entries and automating the process for multiple industries.
- Machine Learning – Deep Learning
- Designed the solution using one of the most powerful supervised deep learning technique – Convolution Neural Network
- Developed Caffe model using AlexNet architecture and trained using MNIST database of 50,000 images
- Tested the model with 10,000 images
- Developed script in Python to export model parameter and converted using CMSIS-NN library to deploy on edge – i.MX RT platform
- Developed application in C on NXP i.MX RT for capturing the image, pre-processing and executing deep learning model for digit classification
- Analyzed the performance of the model by tweaking different parameters and layers of the AlexNet. architecture
Caffe Model | AlexNetarchitecture | MNIST database | CMSIS-NN library | C | Machine Learning | Supervised Deep Learning
Machine Learning Model Building & Training
- Enabled client to capture new market with application developed on new processor series
- Improved worker’s efficiency with automated digit recognition and classification process