Project
AI4IMAGE
Video and images analysis
What is AI4Image?
AI4Image is a machine learning solution based on different types of neural networks, each specialized in recognizing particular characteristics of the video.
In particular, it can recognize:
- Human figures, regardless of their relative position
to the camera (front, back, side) and of
framed part (full figure, suerior part,
inferior part); - Objects, 90 different categories are classificable,
including the most common types of two- and four-wheeled vehicles; - Flames, using a mixed approach.
The system is able to perform detailed analyzes, for example by searching for the face of a human figure, when this is detected, or by searching for the license plate of a vehicle.
Image pretreatment
AI4Image also performs image pretreatments to facilitate the task
of the artificial intelligence.
Currently, the image enhancement layer includes interventions on:
- Luminosity
- Noise
Web based interface
The AI4Image input is represented by a video file in any of the most common formats on the market, uploaded via a simple web based interface.
The search parameters are entered in the same interface, such as:
- Objects or events to be researched,
with multiple choice - Minimum confidence threshold of
detection
Output
The output is represented by video frames containing the sought evidence.
Each detection is returned the confidence level.
A tabular output of the surveys is also provided, containing information such as time stamp, detected category and confidence interval, as well as all information regarding the processing carried out.
Software as a Service
AI4Image is proposed as a software-as-a-service, in full compliance with current legislation on privacy, data management and chain of custody of the video record.
Request an online demo
One of our consultants will assist you with the explanation.