High-Resolution Litchi Fruit Image Dataset for Maturity Detection

1) The demand for high-quality, fresh fruits is universal. In today’s health-conscious society, people are increasingly selective about their food choices, believing that consuming spoiled fruits can negatively impact their health. As a result, the fruit market often suffers, leading to significant economic losses. A major contributor to fruit spoilage in Bangladesh is the manual method used to assess fruit maturity. If fruits aren’t harvested at the right time, they can deteriorate quickly. Therefore, accurately distinguishing between ripe and unripe fruits is crucial for determining the optimal harvest period. Litchi, a highly nutritious fruit and a key crop in Bangladesh, faces substantial daily financial losses due to spoilage. This highlights the urgent need for an automated system to categorize litchi fruits as mature, and immature which would be invaluable to farmers, vendors, and the fruit processing industry. (2) In the modern era, computer vision techniques have shown great potential in handling tasks like classification and detection. (3) To support the development of computer vision-based algorithms, we introduce a comprehensive dataset for litchi fruit that includes Maturity Detection datasets. The Maturity Detection Dataset classifies fruits as either immature or mature, This classifications were made in consultation with an agricultural expert from a specialized institute. (4) The dataset includes 3,400 images of mature, immature litchi fruits collected from demonstration areas across three locations in Bangladesh. Additionally, 20,000 augmented images were generated using flipping, width, height shifting, brightening, rotation, shearing, and zooming to expand the dataset.1) 高品质新鲜水果的需求具有普遍性。在当下健康意识觉醒的社会中，人们对食品选择愈发挑剔，认为食用变质水果会对健康产生负面影响。这一现状往往给水果市场带来冲击，造成显著的经济损失。在孟加拉国，人工评估水果成熟度的方法是导致水果腐败的一大主因。若未能在适宜时机采收果实，水果会快速腐坏变质。因此，精准区分成熟与未成熟果实，对于确定最佳采收期至关重要。荔枝作为营养丰富的水果，同时也是孟加拉国的核心作物之一，因腐败问题每日面临巨额经济损失。这凸显出开发自动化分类系统以区分荔枝果实成熟度的迫切需求，该系统将对种植户、经销商及水果加工行业具有极高应用价值。 (2) 当今时代，计算机视觉（Computer Vision）技术在分类、检测等任务中展现出巨大应用潜力。 (3) 为支撑基于计算机视觉的算法开发工作，我们构建了一套涵盖成熟度检测数据集（Maturity Detection Dataset）的荔枝综合数据集。该成熟度检测数据集将荔枝果实划分为未成熟与成熟两类，分类结果均经来自专业农业研究机构的农业专家审定。 (4) 本数据集包含从孟加拉国3个示范采集区域收集的3400张成熟与未成熟荔枝果实图像。此外，通过翻转、宽平移、高平移、亮度调整、旋转、错切及缩放等数据增强操作生成了20000张增强图像，以扩充数据集规模。