Context People can immediately and precisely identify that an image contains 1, 2, 3 or 4 items by a simple glance. This phenomenon is known as Subitizing. An interesting data problem is to predict the existence and number of salient objects in a scene using holistic cues. The phenomenon, known as Subitizing, inspires us to pursue the task of Salient Object Subitizing, i.e. predicting the existence and the number of salient objects in a scene using holistic cues. Original data source and paper are by researchers at the Computer Science division of Boston University (specific reference can be found below in Acknowledgment section). Content Total of 1224 images originating from four public image datasets: COCO, VOC07, ImageNet and SUN. Each image is labeled as containing 0, 1, 2, 3 or 4+ salient objects by Amazon Mechanic Turk workers. This labelling information along with bounding box annotations can be found in **imgIdx.mat**, a matlab structure array that stores information about the images. The MSO dataset are a subset from the test set of the SOS dataset. The team at Boston University removed some images with severely overlapping salient objects or are ambiguous for labelling the indicated number of salient objects. As such, only 1224 images remain, out of 1380 images from the SOS test set. More than half of the images in the MSO dataset contain either zero salient objects or multiple salient objects. This is designed to provide a more realistic setting to evaluate salient object detection methods. - 0 salient objects: 338 images - 1 salient object: 611 images - 2 salient objects: 155 images - 3 salient objects: 100 images - 4+ salient objects: 20 images Acknowledgements Original data source and banner image: http://cs-people.bu.edu/jmzhang/sos.html Users of this dataset are asked to cite the paper: Jianming Zhang, Shugao Ma, Mehrnoosh Sameki, Stan Sclaroff, Margrit Betke, Zhe Lin, Xiaohui Shen, Brian Price and Radomír Měch. "Salient Object Subitizing." To appear in Proc. IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2015. Inspiration - Which images contain 2 objects? What about 3 objects? [1]: https://www.kaggle.com/jessicali9530/mso-dataset/data/image_objectnumdist.jpg