Neural Network
- the first stage: dataset - a multiple set of data, on which basis the independent training of the neural network takes place
- after successful training, the neural network is embedded in the usual video analytics
- the neural network analyzes hundreds of frames, spending on this task thousands of times less time than a person
- detection is carried out in real time - the operator observes all predetermined indicators on the screen and receives urgent notifications of trigger situations
The neural network processes the data using the power of the drone's onboard CPU. This means that the drone automatically analyzes the information during the flight and immediately sends post processed footage.
Obtaining up-to-date information is carried out in one step - there is no waste of precious time for completing the flight mission, returning to the droneport, unloading.
Obtaining up-to-date information is carried out in one step - there is no waste of precious time for completing the flight mission, returning to the droneport, unloading.
Implementation cases
Helmets & и PPE
helmets, personal protective equipment and other necessary equipment wearing monitoring
Number of personnel and machinery
determination of the number of personnel and transport:
type, cargo in bodies, operational/ non-operational state
type, cargo in bodies, operational/ non-operational state
Defectoscopy
infrastructure maintenance and detection of corrosion, blistering, pits, mesh cracks, leaks
License plates identification
text recognition, contractor’s logos identification
Deviant behavior
deviant behavior monitoring:
fights, vandalism, penetration of protected area
fights, vandalism, penetration of protected area
Animal monitoring
tracking the population of rare species, monitoring animals in pastures, control during the hunting season
Harvest estimation
plants drying monitoring , stages of fruit ripening, drawk spread monitoring
Fire hazards
monitoring of protected areas, park and forests and hazardous areas, monitoring of infrastructure
The types of input data can be different: photo, video, symbols, text, sound, thermal images, chemical reagents, spatial indicators.
The possibilities of a neural network for defining typical cases are practically endless. After training, the neural network can permanently close the largest tasks and, according to the sequence of values, predetermine the future.
The possibilities of a neural network for defining typical cases are practically endless. After training, the neural network can permanently close the largest tasks and, according to the sequence of values, predetermine the future.