Video Analysis Product (Beta version)
DeepFire
OVERVIEW
Automatically detects anomalies or signs of abnormal conditions by accurately analyzing the state of a system which undergoes complex time-series transitions such as "combustion," “viscosity”, “liquid”, etc.
Based on the video, the AI automatically assesses and classifies the state of the system, with comparable results to what was previously performed visually by humans.
Camera video AI monitoring makes it possible to detect anomalies, minimize human load, increase operational performance, optimize performance, and detect equipment changes.
FEATURES
-
Analyzes time-series changes in video information such as "combustion," "viscosity," and "liquid" to quantify anomaly scores and classify scenes.
-
Easy to prepare data for model training
AI training is possible using only videos of normal conditions (only for anomaly detection models).
-
Provide optimal systems to meet business needs
Instead of providing a generic solution, we design and build the optimal system configuration to meet your business needs and processes.
BEFORE AND AFTER
Before
Monitoring and storing data on combustion conditions using cameras, but not being able to utilize it effectively.
- It is difficult for an operator to keep an eye on the monitor 24 hours a day, 365 days a year.
- As a result, video monitoring is used only in a limited way, for example, to review the cause of a problem when it occurs.
Difficult to extract useful information from video analysis and unable to utilize it
- Traditional analysis techniques (e.g., luminance distribution) are unable to extract effective information from complex state changes.
- More difficult to feed into a control system and analyze than numerical data from sensors.
After using DeepFire
By simply collecting videos of normal conditions and letting AI learn from them, it is possible to constantly monitor and alert on abnormal situations!
- No complex feature engineering or programming is required to define abnormal conditions.
By utilizing both sensor data and video, the equipment utilization rate is also improved
- Previously unnoticed signs can be seen in the video before they are detected by the sensors.
- Comprehensive use of video combustion and sensor information to enable more advanced control and operation.
CASE
-
01
Automatic combustion state monitoring.
-
02
Visualization of the state of stirring and mixing of "viscous" objects in food manufacturing processes.
-
03
Automated monitoring of liquid processing status.
-
04
Anomaly detection of conveyor equipment in manufacturing plants.
FLOW
We propose and develop the most appropriate combination of technologies, with AI technology as an option, to solve issues that meet your business needs.
The duration and content of each phase should be discussed according to the issues. Please feel free to contact us.