This post describes the problem of automatically detecting people in thermal and thermal images. Thermal video is recorded in a meadow with a small forest with up to 3 people at various positions and distances from the camera. All videos are recorded at night, but in a variety of weather conditions, fine weather, rain, and fog to simulate the realistic conditions that can occur when monitoring and monitoring protected areas. increase. Presents human detection results in a custom dataset of thermal video using a ready-to-use YOLO convolutional neural network and a YOLO network trained on a subset of the dataset. YOLO is a pre-trained object detector with COCO image datasets of RGB images of various object classes. The results of the test experiments show that the performance of person detection in thermal images is significantly improved in terms of the average accuracy of the trained YOLO model over the original model.
suntech software solutions provides best switch to on using Raspberry project with complete description and project report, and complete on call guidance related to the project order now.
Component List
- raspberry pi
- Webcam
- Power supply
- wires and jumper
- AMG8833 IR 8x8 Thermal Imager Array Temperature Sensor Module
Features
- Thermal vision of human or any substance or object comes in front of camera
