地球资源数据云——数据资源详情

苹果或西红柿 - 图像分类

该数据集《Apples or tomatoes - image classification》主要用于多分类任务，数据形态以图像为主，应用场景偏向农业场景。 题目说明：Build an image classification model 任务类型：图像多分类。 建议流程：先检查类别分布与脏样本，再用迁移学习（如 ResNet/EfficientNet）建立基线。 评估建议：使用分层切分或交叉验证，优先关注 F1、Recall、AUC 等分类指标。 可用文件：未检测到标准 CSV，可优先查看目录中的索引或说明文件。 Images of tomatoes and apples collected via web scraping. These two fruits have been chosen because they look similar and so should be challenging classify. Images of red, yellow and green apples/tomatoes have been split into a train and test set. Image sizes usually range between 100x100 pixels and 300x300 pixels. Since the train set is small, you might want to consider using transfer learning to build the best model.