This dataset, named CWF-788 (CWF is short for crop in weedy field), contains 788 images of transplanted cauliflowers in weedy fields along with high-quality labels. The images have been captured on May 23rd 2018 (sunny) and June 13th 2018 (cloudy) in two different plots of fields in Tongzhou, Beijing. The cauliflower plants have respectively grown for approximately 7 and 5 weeks since being transplanted into the fields. The plants have grown slowly in the early stage due to low temperatures. No herbicide has been applied to either of the fields. The images captured on May 23rd have partial shadows cast on the leaves of the plants and weeds. A digital camera (Canon PowerShot SX150 IS) and two smart phones (iPhone 6 and Huawei Note 8) have been used for image acquisition. The resolutions of the images output from those devices are 1200×1600, 2448×3264, and 4160×3120, respectively. The crop region in each label image has been finely extracted with more than 400 points manually selected on its contour. The 788 images in CWF-788 are divided into three parts: 400, 88, and 300 images are used for training, validation and testing, respectively. There is no overlap between any two subsets. The resolution of this data set are 400×300 and 512x384. Please consider to cite the related article: Nan Li, Xiaoguang Zhang, Chunlong Zhang, Huiwen Guo, Zhe Sun, and Xinyu Wu, “Real-time crop recognition in transplanted fields with prominent weed growth: a visual-attention-based approach,” IEEE Access, 2019, 7(1): 185310-185321, DOI: 10.1109/ACCESS.2019.2942158

The original repository of this dataset is in GitHub, at https://github.com/ZhangXG001/Real-Time-Crop-Recognition We have published the data into zenodo to preserve from accidental removal.