Avinton Japan (Avinton) specializes in analyzing big data using machine tools such as Python and TensorFlow, and creating machine learning mechanisms using it.
We are developing solutions that gather various data such as images, sounds and numerical values, analyze them, and improve the efficiency of work and improvement.

Some semiconductor manufacturers stood in the production line of IC chips and visually checked them.
Craftsmen are going to check each and every one, relying on long-standing experiences of flowing IC chips one after another.
For the missing item inspection, a skilled technician is required, training has become difficult, and it was a task of wanting to automate.

Approximately 3,000 - 4,000 chips are embedded in a single semiconductor.
We provide training data from users, and make big data.
Based on that, image recognition / missing item confirmation is carried out, and the data is further collected.
The iteration becomes machine learning, and analysis with higher accuracy becomes possible.

By the above, we realized chefs' training costs, labor cost reduction and work efficiency.

In addition to analyzing the image processing of MRI data for medical use and image processing of surveillance cameras,
It was actually developed for users and is still being utilized.

For companies that used SAP and ERP to conduct sales and sales forecasts,
We received data at the date and time of the past few years, and based on that data we automate sales forecasts.
Reduction of personnel expenses and improvement of work efficiency are realized.

Analysis and learning become possible by converting speech into numerical data (decibels).
In the noise measuring company, we collect sound collectors on the truck, run nationwide, and collect environmental noises.
We divide it into trucks, motorcycles, construction, etc. and classify them into levels, and make them heat map.
Until now, people have been doing, but building base based on audio data for analysis.
By digitizing and judging the sound, it is possible to create a heat map in the same way.
By introducing machine learning also, this leads to improvement of accuracy.