We created a stereo dislocation dataset composed of image sequences of three different materials, representing large diversity of dislocations shapes and contrasts.
Bright-field images of dislocations were acquired from TiAl alloy, GaN sample, and high-entropy alloy using a transmission electron microscope in scanning mode (STEM).
Details of imaging conditions and specimen preparation will be found in our paper.
Our dislocations dataset consists of 90 labeled stereo pairs greyscale images split into train, validation and two test subsets.
Segmentation masks and disparity maps were manually labeled for the task.
Download dataset [28 MiB]
Example frame:
