Oktoberfest Food Dataset

我们发布了一个真实、多样且具有挑战性的图像目标检测数据集。该数据集在德国的一个啤酒帐篷中录制，包含15种不同的食品和饮料类别。我们手工创建了超过2,500个对象标注，涵盖了通过视频摄像头在收银台上方捕捉的1,110张图像。此外，我们还提供了包含数天视频的剩余600GB（未标注）数据。同时，我们提供了训练好的模型作为基准。可能的应用包括能显著加快流程的自动收银系统。

We present a real-world, diverse and challenging image object detection dataset. This dataset was collected in a beer tent in Germany and includes 15 distinct food and beverage categories. We manually created over 2,500 object annotations for 1,110 images captured by a video camera mounted above the checkout counter. Additionally, we provide the remaining 600GB of unannotated data containing several days of video recordings. We also release a pre-trained model as a benchmark for this dataset. Potential applications include automated checkout systems that can significantly accelerate the checkout workflow.