Ask Your Question

Revision history [back]

Common approach: Berkeley Segmentation Dataset (BSDS300) [1] using the precision-recall framework introduced in [2]. Taken from [3].

The BSDS300 consists of 200 training and 100 test images, each with multiple ground-truth segmentations.

An extension of the BSDS300 is created (BSDS500) [3]. It is is an extension of the BSDS300, where the original 300 images are used for training / validation and 200 fresh images, together with human annotations, are added for testing. Each image was segmented by five different subjects on average. Performance is evaluated by measuring Precision / Recall on detected boundaries and three additional region-based metrics.

Link to download BSDS500 :

See also this publication: 'Benchmarking Image Segmentation Algorithms' [4] And these links:

[1] D. Martin, C. Fowlkes, D. Tal, and J. Malik, “A database of human segmented natural images and its application to evaluating segmentation algorithms and measuring ecological statistics,” ICCV, 2001. (

[2] D. Martin, C. Fowlkes, and J. Malik, “Learning to detect natural image boundaries using local brightness, color and texture cues,” PAMI, 2004. (

[3] Arbelaez, P., Maire, M., Fowlkes, C., & Malik, J. (2011). Contour detection and hierarchical image segmentation. Pattern Analysis and Machine Intelligence, IEEE Transactions on, 33(5), 898-916. (

[4] Estrada, F. J., & Jepson, A. D. (2009). Benchmarking image segmentation algorithms. International Journal of Computer Vision, 85(2), 167-181. (