Datasets > Modality > Images > DispScenes

The DispScenes dataset was created to address the specific problem of disparate image matching. The image pairs in all the datasets exhibit high levels of variation in illumination and viewpoint and also contain instances of occlusion. The DispScenes dataset provides manual ground truth keypoint correspondences for all images.

Source: Matching Disparate Image Pairs Using Shape-Aware ConvNets

License

  • Unknown

Modalities

Languages

Tasks