10 papers
Efficient Modelling Across Time of Human Actions and Interactions
Alexandros Stergiou
This thesis focuses on video understanding for human action and interaction recognition. We start by identifying the main challenges related to action recognition from videos and r…
The Mind's Eye: Visualizing Class-Agnostic Features of CNNs
Alexandros Stergiou
Visual interpretability of Convolutional Neural Networks (CNNs) has gained significant popularity because of the great challenges that CNN complexity imposes to understanding their…
Refining activation downsampling with SoftPool
Alexandros Stergiou, Ronald Poppe, Grigorios Kalliatakis
Convolutional Neural Networks (CNNs) use pooling to decrease the size of activation maps. This process is crucial to increase the receptive fields and to reduce computational requi…
Multi-Temporal Convolutions for Human Action Recognition in Videos
Alexandros Stergiou, Ronald Poppe
Effective extraction of temporal patterns is crucial for the recognition of temporally varying actions in video. We argue that the fixed-sized spatio-temporal convolution kernels u…
Learn to cycle: Time-consistent feature discovery for action recognition
Alexandros Stergiou, Ronald Poppe
Generalizing over temporal variations is a prerequisite for effective action recognition in videos. Despite significant advances in deep neural networks, it remains a challenge to…
Learning Class Regularized Features for Action Recognition
Alexandros Stergiou, Ronald Poppe, Remco C. Veltkamp
Training Deep Convolutional Neural Networks (CNNs) is based on the notion of using multiple kernels and non-linearities in their subsequent activations to extract useful features.…