PETS 2015
Karlsruhe, Germany - August 25 2015

In Conjunction with AVSS 2015

 

PETS 2015 Challenge uses two datasets that are ARENA and P5 datasets. ARENA dataset was used in its full form in PETS 2014 Challenge; here we use a more restricted set of ARENA dataset by including few scenarios that are more relevant to PETS 2015 Challenge. The selected scenarios from ARENA and P5 datasets are grouped into 'Normal', 'Warning', 'Alarm' categories. 'Normal' alludes to activities that do not pose any threat. 'Warning' refers to abnormal activities that may potentially develop into a threat. 'Alarm' refers to activities that cause a threat in the scene and hence require immediate action. Below is given a description of the two datasets.

 

ARENA Dataset

Overview


The dataset contains sequences with different activities around a parked vehicle in a parking lot.

Note: this dataset was recorded within the framework of the EU project ARENA, whose acronym stands for 'Architecture for the REcognition of threats to mobile assets using Networks of multiple Affordable sensors'.

The recorded (acted) scenarios contain abnormal behaviour around the parked vehicle. The dataset was recorded using the environmental CCTV cameras (that allow for the surveillance of the area around the vehicle) and the visual (RGB) cameras mounted on the vehicle itself. For more details about the recording characteristics of ARENA dataset, please visit this page.

Scenarios


 
List of scenarios from ARENA dataset

Scenario type

ID Description Summary video*
Normal N1_Arena Three persons walking in a group, pass the truck, and split.

Warning W1_Arena Driver falls after being pushed by someone running.

Alarm A1_Arena Driver involved in a fight with someone. Two more people come to hit driver and brought to ground.

A2_Arena Driver attacked from someone from a car and brought to floor.

* Note that the above-listed summary videos present a synchronized coverage from multiple camera views only to provide an overall scene visualization. The data for the Challenge includes the single-camera video sequences extracted from the camera views of the above tabulated scenarios as given in 'Authors instructions' page.

 

 

P5 Dataset

Overview


The dataset contains sequences with different activities staged at the OKG nuclear plant outside Oskarshamn, Sweden.

Note: this dataset was recorded within the framework of the EU project P5, whose acronym stands for 'Privacy Preserving Perimeter Protection Project'.

The dataset was recorded by multiple types of surveillance sensors including digital IP cameras and the thermal sensors. See the section below for sensor specications and location information.

Recording characteristics


The sensor properties are summarized in the table below.

Name Model Resolution (pixels)
VS_1 Basler BIP2-1300x-dn 1280 x 960
VS_2 Basler BIP2-1300x-dn 1280 x 960
VS_3 Basler BIP2-1300x-dn 1280 x 960
TH_1 FLIR SC655 640 x 480
TH_2 FLIR SC655 640 x 480
TH_3 FLIR SC655 640 x 480
TH_4 FLIR A65 640 x 512

 

Scenarios


List of scenarios from P5 dataset

Scenario type

ID Description Summary video*
Normal N1_P5 A vehicle driving across the scene.

Warning W1_P5
  • A group of 6 people walking across the scene.
  • 1 person starts to run, the remaining 5 people split into 2 smaller groups later.

Alarm A1_P5
  • A boat approaching the bank.
  • The person throwing a bag from the boat.
  • The boat driving away from the bank.
  • A person appears from the woods to fetch the bag, then disappears into the woods.

* Note that the above-listed summary videos present a synchronized coverage from multiple camera views only to provide an overall scene visualization. The data for the Challenge includes the single-camera video sequences extracted from the camera views of the above tabulated scenarios as given in 'Authors instructions' page.



Download Information

To download the video sequences, please fill out and submit the User Agreement form here. Following the form submission, you will receive an email within 24 hours with the information to download the dataset.

Please cite the following paper (in your publications, presentations etc.) if you use PETS 2015 Datasets:

 

Legal note: The video sequences are copyright University of Reading and permission is hereby granted for free download for the purposes of the PETS 2015 workshop and academic and industrial research. Where the data is disseminated (e.g. publications, presentations) the source should be acknowledged.