Datasets
Standard Dataset
OSR_DATASET
- Citation Author(s):
- Submitted by:
- Yihan Xiao
- Last updated:
- Fri, 03/14/2025 - 07:20
- DOI:
- 10.21227/fp63-6703
- License:
15 Views- Categories:
- Keywords:
Abstract
This dataset provides electromagnetic spectrum feature data for target recognition in combat formations, supporting both closed and open set scenarios. It includes three subsets: a closed set with known target types, open set 1 with one unknown target type, and open set 2 with multiple unknown target types. Each dataset contains extracted target features, adjacency matrices representing communication links, and ground truth labels. The dataset covers radar and communication attributes, including carrier frequency, pulse characteristics, modulation types, power, and movement parameters. Preprocessing methods such as min-max normalization and label encoding are applied. The adjacency matrix captures target relationships, making the dataset suitable for graph-based learning models. This dataset facilitates research in open set recognition and target classification under real-world conditions of incomplete and unknown target data.
This dataset comprises electromagnetic spectrum feature data extracted from three combat formations: closed set, open set 1, and open set 2. The closed set includes only target types present in the training set, while open set 1 contains one unknown target type, and open set 2 includes multiple unknown target types. For each dataset, the provided information includes target characteristics, an adjacency matrix representing communication links, and the true labels of target types. Electromagnetic spectrum features are processed using min-max normalization or label encoding, covering key parameters such as radar carrier frequency (RCF), pulse width (PW), pulse amplitude (PA), repetitive pulse interval (PRI), inter-pulse modulation type (PM), intra-pulse modulation type (IPM), communication frequency (CCF), bandwidth (BW), power (P), modulation type (CM), movement speed, and altitude. The adjacency matrix defines the communication relationships between targets, offering structural context for graph-based target recognition and classification.
