Application of deep learning-based techniques for performing live sorting on aerial thermal camera images

Artificial intelligence is currently used in many applications and research areas. Artificial intelligence is supported in everything that will automate the work that requires manpower, communicate, understand images and conversations, in short, make people's lives easier. For this reason, artificial intelligence techniques that can make systems more intelligent are also developing rapidly. One of the systems where artificial intelligence is used is imaging systems. The data taken from various camera systems have become processable and interpretable with the image processing techniques of artificial intelligence. There are many image retrieval systems available today. Them; installed energy X-rays for the different structures of the rotation of the image Area X-ray devices, ultrasound devices that can generate images through sound waves, radio frequency, electromagnetic force who obtained the image with MR devices, using radar precious minerals in the ground, water supply, underground spaces that can display various data, such as 2D or 3D detectors, capable of imaging with infrared rays, thermal imaging systems are examples. Thermal imaging systems are imaging systems in which the temperature differences of objects are colorized using the values of the infrared ray. As an imaging method, it is based on IR (infrared) energy contained in the electromagnetic spectrum. Nowadays, thermal imagers are used in many different fields such as health, agriculture, and construction, especially in the defense industry. With the developing technology, information about various environments can be collected and processed through thermal imaging systems and artificial intelligence. In this study, research has been conducted on where and how artificial intelligence techniques are applied on thermal imaging. In the study, many different studies such as the use of thermal 3D modeling object buildings made with the help of thermal cameras, motion and target detection, alcohol individual detection, determination of structural problems in agricultural and forestry work sites were included. In this thesis, deep learning-based techniques were used to classify the objects in thermal camera images autonomously. Within the scope of the study, 836 images consisting of 4 classes were used. the images were classified using the K nearest neighbor algorithm and the Support vector Machines algorithm, which are integrated into basic deep learning architectures. Keywords: Thermal Camera, Infrared Ray, Deep Learning, Machine Learning