Unimore logo AImageLab
Back to the research area

INSECTT Anomaly detection SETA Dataset

Long, untrimmed real-world surveillance videos with 11 realistic anomalies recorded with a series of CCTV cameras placed inside SETA buses.

Dataset Hightlights:

  • 58 different bus rides from 2020.
  • up to 5 cameras with multiple angles
  • 31 hours of video
  • Divided in 182 sequence

The dateaset is up to date the largest dataset in the world about public transports anomalies in a single setting(Scenario).  It uses non-crowdsourced data. Data are acquired on field in the operative scenario.



In order to be relased and compliant with privacy regulations the dataset has been anonimized with a custom software developed in the WP2 of INSECTT project

In brief the anonymization pipeline performs the following operations:

  1. Grayscale conversion.
  2. Downscale of the resolution by a factor.
  3. Upscale to the original resolution.

Dataset composition on anomalous event on buses:

The entire dataset is made up of 349 videos, of which 255 normal and 92 anomalous.

  • The dataset collects 58 different bus rides from 2020.
  • up to 5 cameras recording different sections of the bus cabin
  • multiple angles,
  • 31 hours of video
  • Divided in 182 sequences splitted in 349 clips.

Train Test split:

For the train set, we used 50% normal videos (126) and 50% abnormal videos (45), due to the low number of abnormal videos. The test set, on the other hand, consists of the remaining 50% of anomalous videos and the same number of normal videos, to achieve class balancing on test data .

 

Availability:

Dataset is available for research purposes only and upon request. Contact segreteria.aimagelabATunimore.it for requests.