Remote Sensing Agreement Dataset
The results of an experiment in which 14 experts annotated a remote sensing image, intended for the study of annotator agreement and ‘ground-truth’ estimation algorithms (reference data learning). Included in the Agreement Analysis Toolbox (see the example_case_study subdirectory and the toolbox's readme)
To facilitate and encourage the fair comparison of research we are making a number of passive sonar spectrogram images derived from time-series data publicly available. These spectrograms are generated from recordings of acoustic energy radiated from propeller and engine machinery in underwater sea recordings.
This data is available for non-commercial use; you may freely download the data set for your own exclusive research purposes. All the data contained here has been kindly supplied by Jim Nicholson, QinetiQ at Winfrith.
Should you publish results achieved using these data sets we kindly ask you to cite the following paper:
- T. Lampert and S. O’Keefe, 'On the Detection of Tracks in Spectrogram Images', Pattern Recognition 46 (5): 1396–1408, 2013. (link, pdf)
Real Small Boat Data
The data set is intended for use in testing spectrogram track detection algorithms; the data, matlab functions, and data description are available here.
This dataset contains 4142 spectrograms that have a variety of signal-to-noise ratios and track appearances (frequency modulation and sound source profiles) generated from simulated data. A datasheet describing the parameters used to generate this dataset is available here, and the dataset itself is available here.