Perun Gaze

Video Object Detections Service

Project description

Perun Gaze is a service implemented on the surveillance platform. This service helps administrators to identify third-party objects and people in guarded areas using surveillance cameras with infrared
or thermal sensors.
ANALYSE

Objective

Develop a service that helps to identify detected objects more accurately in an existing video.

The main task was implementing the labeling without changing the initial video in the database.
INNOVATIVE TECHNOLOGIES

Challenges

Ready-made videos with detection objects were taken as a basis. Then, data with coordinates and types of detection objects were calculated separately with the help of Machine Learning (ML).

The result is a visual representation of moving objects in the bounding boxes with the type and percentage of recognition.

Outcome

8
weeks of development
6
developers
10
technologies
Technology Stack
Python
Docker
Rabbitmq
Pytorch
Nvidia
FastAPI
Labelbox
Postgres
Opencv
Wireguard