MF-ID

This is a part of the Cityintrusion-Multicategory dataset for testing and training the network. This dataset contains 2502 training images and 429 validation images. Because our task is a joint task of segmentation and detection. Therefore, we provide the two different sub-dataset for segmentation and detection, respectively. In the seg folder, we provide the original images for training and validation. Besides, the corresponding labels also are provided. Training and validation have 2502 and 429, respectively. In the det folder, we also provide the original images and labels, as the same as seg, training and validation have 2502 and 429, respectively. it is worth noting that our det sub-dataset is in YOLO format to accommodate the needs of our MF-YOLOV5 framework. The full dataset can be downloaded from this website: https://pan.baidu.com/s/1IK3sl-yXa8TVcJ9TxJajrA?pwd=u7c4 and our code can be found in this website: https://github.com/1012537710/MF-ID.