Earth observation (EO) provides various multi-platform, multi-temporal, and multiresolution remote sensing imagery for dynamic monitoring of planet Earth, with a wide variety of uses. Crop monitoring is a typical application, which involves timely gathering of the information of crop types, boundaries, and dynamic changes during the whole crop growth period. However, most of the existing datasets and benchmarks focus on medium-resolution (≥ 10 m) classification of the main crop type by using satellite image time series (SITS), where the individual boundaries (parcels) and the dynamic changes of the crop cannot be obtained, due to the limited spatial resolution and the lack of multi-season annotation. In this paper, a multi-platform, multi-temporal, and multi-resolution (M3) remote sensing crop segmentation dataset (M3CropSeg) is introduced for very high resolution (VHR, 1 m) crop semantic segmentation to instance segmentation and dynamic segmentation. Specifically, M3CropSeg contains 16311 pairs of airborne VHR (1 m) and SITS (10 m) images, with 45 crop types and 101k instance annotations, covering a 26,000 km2 area of California in the U.S. M3CropSeg has various challenges, including M3 data fusion, class imbalance, fine-grained classification, and multilabel classification. Three tracks are designed for M3CropSeg, i.e., M3 semantic segmentation, M3 instance segmentation, and M3 dynamic segmentation, to obtain high-resolution pixel-level, parcel-level, and multi-season crop types, respectively. The corresponding benchmarks are also provided to address the above challenges, along with a variety of experimental analyses.