We don't know when an accident like this may happen, such as a person falling in an unnoticeable location, causing property damage at night, or causing violence in the area. In many circumstances, security guards have a limited time for keeping an eye on the facility, and it is too late to locate it.
With the implementation of Asilla's product, security will be strengthened by utilizing footage from security cameras 24 hours a day, seven days a week. A security camera continually monitors a facility on behalf of a person, alerts guards when something wrong happens, and urges them to take immediate action.
Asilla has developed a proprietary algorithm for tracking persons in time-series images. It is possible to examine the individual identification ID even if it is hidden for a short period by improving tracking performance.
Each device has succeeded in generating an AI model through an independent experience according to each space by having an autonomous learning function within the AI device. Individual devices can therefore act as a substitute to a section of the cerebral cortex (or the fusiform gyrus area) that is suited for the space, allowing for very precise identification of body parts. As AI learns and executes autonomously, it is able to improve accuracy without the assistance of humans.
For processing camera images, open source such as OpenPose has a high computation cost, and if real-time performance is required, the introduction cost will be significant. Asilla was able to transfer information from a huge high-precision model to a compact one, resulting in a considerable weight reduction. One server can process around 80 cameras at the same time using this method.
Detect behaviorsPatented
AI models optimized for the angle of view are automatically generated by self-learning. It is also feasible to prevent accidents by recognizing "abnormal behavior" that deviates from normal behavior.
For each camera | 5,000Yen/ monthVolume discount available |
---|---|
・Initial cost 0 Yen ・Consistent support from equipment selection, installation, and operation |
"Behavior identification system"Patent number:6793383 | Tracking the same person with mulpiple cameras |
---|---|
"Behavior estimation device"Patent number:6525181 | Algorithm detects the backgroung and adjusts the possibility of behavior |
"Behabior estimation device"Patent number:6525179 | ALgorithm recognizes behavior from joint coordinate values |
"Target number identification device"Patent number:6525180 | Algorithm tracks people in the area in chronological order |
"Suspicious/abnormal subject detection device"Patent number:6647489 | Detects abnormal behavior of a person in area |
"Abnormal behavior detection device"Patent number:6692086 | Detects abnormal behavior from the behavior of surrounding people |
"Suspicious/abnormal subject detection device"Publication number:2020-091856 | Detects abnormal behavior from the behavior of surrounding people |
For product inquiries and consultations