Recent advancements in imaging, electronics, and computer science have engendered significant progress in non-destructive testing and quality monitoring within the agro-food industry. This progress is particularly evident in integrating infrared thermal imaging (TI) and artificial intelligence (AI) techniques. As a non-contact method, AI-based TI holds promise in detecting various quality attributes and has found extensive applications in agriculture, food processing, and post-harvest fruit handling. This paper delves into recent applications of AI-based thermal imaging, specifically in post-harvest fruit handling. The introduction provides a comprehensive overview of the challenges faced in the post-harvest fruit handling industry while emphasizing the advantages of AI-driven thermal imaging technology. The detailed thermal imaging system encompasses both passive and active thermography techniques. This paper provides an in-depth exploration of artificial intelligence, focusing on machine learning and deep learning. It highlights the significance of convolutional neural networks (CNNs) and their architectural phases. Subsequently, critical applications of AI-based thermal imaging in post-harvest fruit quality assessment are discussed. These applications encompass bruise detection, maturity identification, condition monitoring, grading and sorting, pest and disease detection, and considerations for packaging and supply chain management. Furthermore, this paper addresses the challenges and limitations of AI-based thermal imaging in post-harvest fruit handling. In conclusion, this paper discusses future trends in AI-based thermal imaging, emphasizing the potential for increased automation and integration with emerging technologies in the post-harvest fruit handling sector. The insights provided contribute to the ongoing dialog surrounding optimizing quality assessment processes in the agro-food industry.

Guava is a fruit grown predominantly in the tropical and subtropical regions of the world. Its skin is thin and soft. Although postharvest handling and transportation of fruits are unavoidable, hand harvesting is preferred to avoid physical injuries. Visually identifying internal damage is a laborious and time-consuming task. In this study, 4129 thermal and digital images of guava fruits were collected. A digital image-based convolution neural network (CNN) model was developed to classify the quality of damaged and diseased fruits. The surface temperatures of all three levels of maturity-indexed and damaged fruits were measured using thermal images. During the storage period, the temperature of the immature fruits was less than half of the mature fruits. Because the architecture was trained on dehydrated and diseased guava, it accurately predicted fruit quality. The prediction accuracy of the developed CNN model was approximately 99.92%. GRAPHICAL ABSTRACT

Guava (Psidium guajava) is a nutritious fruit known for its origin in Mexico, Central or South America, and the Caribbean. Its production faces declining levels, infections, and disease outbreaks. This dataset focuses on the early identification of guava diseases using image processing and computer vision. Farmers can detect and address diseases promptly by developing an expert system, increasing yields and reducing economic losses. The technology behind this dataset enables sustainable guava farming and disease prevention. This dataset consists of digital and thermal images of guava fruits, including healthy, damaged, and various diseased conditions such as wilt, Anthracnose, canker, and rot. The images are categorized based on the fruit's maturity level (mature, half-mature, and mature) and captured under different drop heights. The dataset also includes information on the damage-inducing methods, storage conditions, image capture schedule, and specific diseases present. The thermal images were acquired using hot air with controlled temperature and velocity.