Document Type : Research Paper

Authors

• Ahmed Abdulsatar Kareem
• Adil Al-Azzawi

Computer Science Department, College of Science, University of Diyala, Diyala, Baqubah, Iraq

Abstract

Due to machine learning-based infrared image and video collections, which enable computers to detect and categorize images with increasing accuracy, the field of image identification has experienced a revolution. This comprehensive research presents an overview of the most current advancements in the infrared image and video collections for computer vision and artificial intelligence. It has largely focused on the infrared picture and video collections that have been collected and categorized for computer vision applications such as object identification, object segmentation and classification, and motion detection. This article covers some of the most well-known machine learning methods, including deep learning, convolutional neural networks, support-vector machines, and decision trees. The basic problems with image identification are examined, and only a few of them include data augmentation, feature extraction, and picture segmentation. We also discuss some recent developments in the area of image identification, including ground-breaking deep learning methods like adversarial training and transfer learning. The discussion ends with a discussion of possible uses and the promise of machine learning for picture identification. Because it analyzes state of the art in machine learning for picture identification in-depth, this survey study is a vital resource for academics and entrepreneurs. We make a distinction between publicly accessible collections and those that are maintained in private, based on the various sensor types, image resolution, size, and research effort within that range. Include a glossary of words, including those for infrared radiation, infrared detectors, and infrared optics, that are crucial to comprehending infrared imaging, along with a description of their applications. This article explores the group's overall statistical relevance from a number of different perspectives. Researchers working in computer vision and artificial intelligence who are interested in managing spectra outside of the optical field might use this survey as a reference.

Keywords

• artificial intelligence
• deep-learning
• convolutional-neural-networks
• learning-based infrared
• machine learning