Datasets
Standard Dataset
AShipClass9
- Citation Author(s):
- Submitted by:
- Haolin Li
- Last updated:
- Mon, 04/07/2025 - 21:48
- DOI:
- 10.21227/8p3v-mc81
- Data Format:
- License:
26 Views- Categories:
- Keywords:
Abstract
This study utilizes the open-source datasets FAIR1M and HRSC2016 as foundational resources to construct an optical remote sensing image dataset for rotated ship target detection. The dataset encompasses nine ship categories: Dry-Cargo-Ship, Engineering-Ship, Fishing-Boat, Motorboat, Tugboat, Passenger-Ship, Warship, Liquid-Cargo-Ship, and Other-Ship. Due to discrepancies in category labels, annotation formats, image dimensions, and storage formats across these datasets, the AShipClass9 dataset was established after undergoing preprocessing steps, including screening, segmentation, format conversion, re-annotation, and merging. AShipClass9 employs the OBB annotation format, where positional information is represented by eight parameters, ${(x_i,y_i ),i \in 1,2,\cdots,4}$. Here, $(x_i,y_i)$ denotes the i-th vertex coordinate of the rectangular bounding box, arranged in a clockwise order. Images are stored in the PNG format. The AShipClass9 dataset consists of 5684 images and 66,493 target instances, divided into training, validation, and test sets in a ratio of approximately 7:2:1.
This study utilizes the open-source datasets FAIR1M and HRSC2016 as foundational resources to construct an optical remote sensing image dataset for rotated ship target detection. The dataset encompasses nine ship categories: Dry-Cargo-Ship, Engineering-Ship, Fishing-Boat, Motorboat, Tugboat, Passenger-Ship, Warship, Liquid-Cargo-Ship, and Other-Ship. Due to discrepancies in category labels, annotation formats, image dimensions, and storage formats across these datasets, the AShipClass9 dataset was established after undergoing preprocessing steps, including screening, segmentation, format conversion, re-annotation, and merging. AShipClass9 employs the OBB annotation format, where positional information is represented by eight parameters, ${(x_i,y_i ),i \in 1,2,\cdots,4}$. Here, $(x_i,y_i)$ denotes the i-th vertex coordinate of the rectangular bounding box, arranged in a clockwise order. Images are stored in the PNG format. The AShipClass9 dataset consists of 5684 images and 66,493 target instances, divided into training, validation, and test sets in a ratio of approximately 7:2:1.
