PETS 2016
Las Vegas, USA - July 1st 2016

In Conjunction with CVPR 2016

PETS 2016 datasets

The ARENA dataset

The ARENA dataset comprises of a series of multi-camera video recordings (22 acted scenarios) where the challenge is the detection and understanding of human behaviour around a parked vehicle employing four visual (RGB) cameras mounted on the four corners of the vehicle itself (see Figure 1). The main objective is to detect and understand the different behaviours discriminating between normal, abnormal/rare behaviour and real threats.

 The dataset has been available for download since PETS2014. Hence, PETS2016 provides the opportunity for researchers and industry to submit methodological advances and results obtained using this data since the 2014 and 2015 workshops.

The ARENA On-board cameras

The on-board camera configuration during the recordings is shown below. Four non-overlapping visual cameras are mounted at each corner of the vehicle. They are represented with light blue colour.

Figure 1: Different sensors and coverage in the ARENA Dataset

The characteristics for the four on-board cameras are as follows:

More details on the ARENA recordings including the Physical localisation and supplementary cameras employed are available at:

The IPATCH dataset

A new IPATCH dataset, collected in April 2015, contains a set of fourteen multi camera recordings (visible, themal) collected off the coast of Brest, France, in collaboration with the AUTOPROTECTION project. The recordings, which represent a series of realistic maritime piracy (abnormal / attack to vessel) scenarios, presents different challenges covering object detection and tracking (fusion of data from sensors with different modalities and sensor handover (tracking objects passing from one FOV to another with minimal overlapping FOV)), event detection and threat recognition. In addition to video data, GPS data is also made available. Altogether, detection, tracking and scene understanding challenges in the maritime domain can now be addressed on this dataset. The data collection was carried out employing the vessel ‘VN Partisan’ (see Figure 2), managed by AUTOPROTECTION project.