MangoDB - A TJC Mango Dataset for Deep-Learning-Based on Classification and Detection in Precision Agriculture

TJC Mango (Mangifera Indica L) was discovered by Tom Ellawala in his mango orchard in Sri Lanka in the early 1990s. The Department of Agriculture recognized it in 2003. This variety is now grown in Sri Lanka's Dry and Intermediate zones. Artificial Intelligence (AI) can assist agriculture producers in making more informed decisions. The study covers prospective mango farming applications of the TJC Mango Dataset (MangoDB) contributed by AI, including yield prediction, disease detection, and fruit quality evaluation. MangoDB is a unique post-harvest TJC Mango fruit image collection. Five TJC mango label classifications have 200 photographs each, totaling 1200 images. Machine learning (ML) models and neural networks trained on the MangoDB can classify and evaluate mango fruit quality, helping Farmers and stakeholders to make better harvesting and post-harvest decisions. A TJC Mango Dataset could boost mango farming's profitability and sustainability with AI. In agriculture, grading significantly will benefit from detailed TJC mango images. Photos were taken with a NIKON D5500 DSLR at ISO 2500, 1/100 sec shutter speed, and 24mm focal length. The photos were in NEF format. Classification algorithms like Fast Forward Neural Network (FFNN), Support Vector Machine (SVM), and K-Nearest Neighbor (KNN) can be used qualitatively to analyze the TJC mango's quality descriptors (colour, surface texture). In addition, MangoDB can be used to implement a rule-based classifier with a combination of ML algorithms to determine the mango's quality status. This study examined TJC mango dataset qualities and compared them to others.