Bicycles parts detection, segmentation and classification
About project
Solution
A neural network has been developed that extracts various parts of the bicycle (steering wheel seat frame, etc.). Different parts are well customized.
Bissines effect
A task
Bicycles can be combined from different parts, so the custom bike algorithm for the client works.
It is necessary to develop a retail tool that allows the client to customize the bike for ordering.
Solution
The main difficulty of the work was that the customer was at a distance - in the United States. Communication was maintained through video communication.
Professional markers were involved to improve the quality of markings
Photographs of various types of bicycles (classic, mountain, electro, etc.) were used.
External Data Collection
1.
Marking CVAT
2.
Neural Network Training
3.
Implementation and monitoring of neural network operation
4.
Development of Telegram bot
5.
Project team
Developer
Dmitry I.
Manager
Frank Sh.
Areas of use
The project is useful for retailers who need to divide various complex goods into parts.
