Segmentation and detection of objects in drone images
A task
It was required to mark images taken from a height of about 50 m using drones, and then train the neural network to recognize territories and PPE (personal protective equipment).
Solution
One of the main difficulties is the processing of a large amount of data. Also, in addition to the segmentation algorithm, there was an introduction of algorithms for recognizing helmets in humans.
Neural Network Training
2.
Implementation and monitoring of neural network operation
3.
External Data Collection
1.
Business effect and areas of use
This decision warns the absence of PPE (personal protective equipment) at the facilities under construction, as well as, using the algorithm, zoning of the territory was carried out. This case is relevant to construction and production companies, as well as useful in the field of agriculture and urban planning.
Project team
Developer
Frank Sh.
