(aside image)

The Content-Based Image and Information Retrieval Group Data Repository

Style and mood ratings, computational features, and content annotations for 50 film clips (2017)


This data set pertains to the following article: Jussi Tarvainen, Jorma Laaksonen and Tapio Takala: "Understanding Quantitative Determinants of Film Mood in Different Types of Scenes", submitted to IEEE Transactions on Multimedia, January 2017.

The data set contains the following data for the 50 film clips used in the study: the clips' timecodes and scene type classifications; human ratings of the clips' style and mood; the values of 14 low- and high-level computational features obtained from the clips; as well as ground-truth data on the clips' shot boundaries, frame-by-frame locations of characters, dialogue contents and timecodes, and music timecodes. Each data file contains a detailed description of its format.

Download data



Style, aesthetics, and affect ratings as well as computational features for 14 film clips (2014)


This data set pertains to the following article: Jussi Tarvainen, Mats Sj√∂berg, Stina Westman, Jorma Laaksonen and Pirkko Oittinen: "Content-Based Prediction of Movie Style, Aesthetics, and Affect: Data Set and Baseline Experiments", IEEE Transactions on Multimedia, vol. 16, no. 8, pp. 2085–2098, DOI: 10.1109/TMM.2014.2357688, December 2014.

The data set contains the following data for the 14 film clips used in the study: human ratings of the clips' stylistic, aesthetic, and affective attributes; and the values of 192 low-level computational features obtained from the clips. Each data file contains a detailed description of its format.

Download data



HUT rector paintings gaze data

Download video data collected with SMI gaze tracking glasses.

With the ObjectDetection (od) software one can detect the paintings and draw heat maps as described in hut-rectors-gaze/HOWTO. Demonstrations of the results can be found as Museum setup gaze tracking demonstrator.