Transfer learning for the classification of video-recorded crowd movements

Mounir Bendali-Braham (1) , Jonathan Weber (1) , Germain Forestier (1) , Lhassane Idoumghar (1) , Pierre-Alain Muller (1)

1 IRIMAS - Institut de Recherche en Informatique Mathématiques Automatique Signal - IRIMAS - UR 7499

The automatic recognition of a crowd movement captured by a CCTV camera can be of considerable help to security forces whose mission is to ensure the safety of people on the public area. In this context, we propose to fine-tune a model from the TwoStream Inflated 3D architecture, pre-trained on the ImageNet and the Kinetics source datasets, to classify video sequences of crowd movements from the Crowd-11 target dataset. The evaluation of our model demonstrates its superiority over the state-of-the-art in terms of classification accuracy.

Mots clés: Video-surveillance; Crowd Behavior Analysis; Convolutional Neural Networks; Transfer Learning

Domaines: Intelligence artificielle [cs.AI]

Mounir Bendali-Braham, Jonathan Weber, Germain Forestier, Lhassane Idoumghar, Pierre-Alain Muller. Transfer learning for the classification of video-recorded crowd movements. International Symposium on Image and Signal Processing and Analysis (ISPA), Sep 2019, Dubrovnik, Croatia. pp.271-276, ⟨10.1109/ISPA.2019.8868704⟩. ⟨hal-03205966⟩