Application scheme-using infrared thermal imager to automatically detect and identify wildlife

Using infrared thermal imager to automatically detect and identify wildlife

In agricultural mowing operations, due to the increase in working width and the speed of agricultural machinery, thousands of wild animals are injured or killed every year, so it is necessary to develop an automatic system that can detect wild animals in crops without stopping the cultivation operation .

In the past two decades, infrared thermal imaging technology has received increasing attention in the research and application of computer vision and digital image processing. Due to the invariability of lighting and the low price of thermal imaging cameras, thermal imaging has become a widely viable technology in outdoor surveillance, pedestrian detection, and agricultural operations. With the emerging technology of drones, it can be used for many detection purposes. And due to the good flight ability and the aircraft equipped with computers and infrared cameras, the UAV can perform advanced and high-precision tasks.

Therefore, the detection and identification of wild animals and plants in farmland is essential for reducing the mortality of wild animals and plants, thereby promoting the friendly breeding of wild animals and plants.

The picture shows the device for capturing visual RGB and thermal image

The work proposed by the researchers contributes to the automatic detection and classification of animals in infrared thermal imaging. The method and results are based on top-view images taken manually from the elevator to inspire people to perform detection and recognition based on drones.

The picture shows the visual RGB and infrared thermal image capture the same scene from 5 m (a), 15 m (b) and 30 m (c)

The method is based on a threshold value dynamically fitted for each frame and a novel feature extraction algorithm, and is unchanged for rotation, scaling, and partial pose. The detected objects are tracked in subsequent images to include time information in the recognition part of the algorithm. The algorithm has been tested in a controlled experiment using real animals in the context of wildlife-friendly farming.

At present, the use of infrared camera and drone technology for automatic detection and identification of wild animals and plants is part of the ongoing research on wild animal and plant-friendly agriculture. Using video taken from actual drones to detect and identify wild animals is also a relatively effective method at present.

Reference materials:

Peter Christiansen, Kim Arild Steen, et al. Automated Detection and Recognition of Wildlife Using Thermal Cameras [J]. Sensors. 14:13778-13793, 2014.