N.B. The owner of this Dataset is The University of Cambridge.

I'm not in any way affiliated with The University of Cambridge. I just thought it would be nice for people to have this dataset available on Kaggle.

Context

The Cambridge-driving Labeled Video Database (CamVid) provides ground truth labels that associate each pixel with one of 32 semantic classes. This dataset is often used in (real-time) semantic segmentation research.

The dataset is split up as follows:

• 367 training pairs
• 101 validation pairs
• 233 test pairs

These splits are also used in many academic papers on semantic segmentation (Brostow et al., 2008b; Sturgess et al., 2009; Badrinarayanan et al., 2017; Yu et al., 2020).

The images and masks for each split are in a separate directory.

Citations

(1) Segmentation and Recognition Using Structure from Motion Point Clouds, ECCV 2008 Brostow, Shotton, Fauqueur, Cipolla

(2) Semantic Object Classes in Video: A High-Definition Ground Truth Database Pattern Recognition Letters Brostow, Fauqueur, Cipolla

The original dataset can be found here: http://mi.eng.cam.ac.uk/research/projects/VideoRec/CamVid

Source / Contact

http://mi.eng.cam.ac.uk/research/projects/VideoRec/CamVid

Image Source

https://sthalles.github.io/deep_segmentation_network