Abstract
Star gooseberry provides immense health benefits and is widely recognized in the Indian medicinal system. It holds significant importance in the food production, pharmaceuticals, and cosmetics industries due to the presence of therapeutic and pharmacological properties. Due to its beneficial properties, gooseberry fruit is widely used in treating various ailments. Therefore, cultivating these fruits presents an opportunity to generate revenue, benefiting both farmers and the agricultural sector. The post-harvest process of fruit typically performs the quality assessment by segregating fruits based on visual characteristics, which is tedious and prone to human error. Hence, there is a need to develop an automated computer vision model to assess the fruit quality more accurately. This study focuses on dataset collection, including image samples of both single and multiple-star gooseberry fruits to automate fruit grading. This dataset has been specifically developed for research purposes, contributing to fruit detection, quality assessment, weight estimation, and classification of fruits at various ripeness stages. Further, it provides researchers with an opportunity to develop an automated system for detecting overlapping fruits and touching contours using machine learning, deep learning, and computer vision systems. Image samples of star gooseberry at different growth stages were collected from orchids in Mysuru, India. The dataset, named "AmlaNet" comprises 792 image samples of star gooseberry, captured against a plain background from varying angles, sizes, brightness levels, and distances. The dataset is organized into four folders such as single star gooseberry fruit, multiple fruits, overlapped, and annotated samples of overlapped star gooseberry fruits including fruit samples with different ripeness stages. This publicly accessible dataset is expected to benefit the research community, enabling advancement in computer vision and AI Applications. It can be accessed at DOI: 10.17632/2255bdy9mm.1.