DIPLECS SWEDEN DATASET 2015 (c) Nicolas Pugeault email: n.pugeault@exeter.ac.uk This dataset contains the data used in articles [1] and [2]. It was recorded by Autoliv Inc. (www.autoliv.com), and provided with permission. The video contains the driver's view from the windscreen for approximately 3 hours of driving. The video is of size 900x244 and greyscale. All frames are labelled according to driving context and driver's actions (see [2] for details). A. The driving context was hand-labelled, with +1 for true and -1 for false. It is stored in the text file "context_data.dat" and the columns labels are listed in the file "context_headers.dat". They record the following categories: * environment (all three mutually exclusive) non-urban inner-urban outer-urban * road type (all three mutually exclusive) single-lane dual-lane motorway * junctions type roundabout crossroads T-junction * other pedestrian-crossing traffic-lights road-markings road-signs B. The driver's actions were recorded from the car's sensors. It is stored in the text file "action_data.dat" and the columns labels are listed in the file "action_headers.dat". They record the following categories: IndexNo: frame number YYYY MM DD date of recording hh mm ss.sss time of the frame capture Clutch Brake AccPed car pedals action Flash [ignore] GazeX GazeY driver's gaze location SkinRes [ignore] BrakeProx AccelProx [ignore] Heading Speed [ignore] HeadRate car steering Accel [ignore] RT90_Xpos RT90_Ypos GPS location REFERENCES: [1] N. Pugeault, R. Bowden, "Learning pre-attentive driving behaviour from holistic visual features", In ECCV 2010, Part VI, LNCS 6316, pp. 154-167, 2010. [3] N. Pugeault, R. Bowden, "How much of driving is pre-attentive?", In IEEE Transactions on Vehicular Technologies, vol. 64, no. 12, pp. 1-15, 2015.