MULTICLOD: Multiclass object detection

Computer vision for smart cars and safer roads

Datasets

KITTI_SEMSEG_UNIZG

  • the dataset is suitable for semantic segmentation experiments
  • we hand-picked a subset of 450 images from the KITTI dataset:
    • the images are semantically annotated into 11 classes
    • annotations feature high quality pixel-level polygonal approximations
  • this is an extension of the work from UAB Barcelona:
    • we improve annotation accuracy of the original 146 images
    • we annotate additional 299 images from scratch
    • we supply dense state-of-the-art disparities for each image
01 01 01

TS2010a

  • around 3300 images with traffic signs at Croatian public roads
  • annotated with:
    • image-wide labels
    • object-level labels and bounding boxes
  • suitable for classification and localization experiments
TS2010a