2016年澳大利亚野外机器人中心果园水果图像数据集

果园水果数据集在三个水果品种（即苹果、芒果和杏仁）果园中收集。苹果树和芒果树的图像使用自动地面车辆拍摄获取；杏仁图像数据使用手持相机获取。苹果、芒果和杏仁水果数据集分别由1120张（308×202像素），1964（500×500像素）和620（308×202 像素）的彩色图像（png格式）组成。上述图像已从原始高分辨率数据中裁剪成小块，以便于训练由于硬件内存限制而禁止使用大图像的深度神经网络。该数据集适用于迁移学习算法的基准测试，也适用于为水果检测和分割开发新的深度学习架构。http://data.acfr.usyd.edu.au/ag/treecrops/2016-multifruit/

The Orchard Fruits Dataset was collected in orchards with three fruit varieties, namely apple, mango and almond. Images of apple and mango trees were captured using an automated ground vehicle, while almond image data were acquired with a handheld camera. The apple, mango and almond fruit datasets respectively comprise 1120 color images (308×202 pixels), 1964 color images (500×500 pixels) and 620 color images (308×202 pixels) in PNG format. All aforementioned images were cropped from original high-resolution data into small patches to facilitate the training of deep neural networks, as large images cannot be used due to hardware memory constraints. This dataset is suitable for benchmarking transfer learning algorithms and developing novel deep learning architectures for fruit detection and segmentation. http://data.acfr.usyd.edu.au/ag/treecrops/2016-multifruit/