Code – Datasets

>Code


 3D Voxel HOG 3D Voxel HOG (3D VHOG) is based on the original Histogram of Oriented Gradients “Pedestrian Detection using Histogram of Oriented Gradients” By Dalal and Triggs. It extends the method by using voxels over pixels and expands the original histograms into 2 dimensions. This implementation was developed for local object structure detection, for use in a risk analysis framework (see Papers Below) in which it is used to classify risk related properties of an object (sharp edges, points etc.). However the feature has also been used successfully in face detection problems.
> Dupre, R. and Argyriou, V. (2015) 3D Voxel HOG and risk estimation. In: 2015 IEEE International Conference on Digital Signal Processing (DSP); 21-24 Jul 2015, Singapore.
> Rob Dupre, Vasileios Argyriou, D. Greenhill, Georgios Tzimiropoulos:  A 3D Scene Analysis Framework and Descriptors for Risk Evaluation. 3DV 2015: 100-108

>Databases

Spontaneous Emotion Multimodal Database (SEM-db) Please send us an email to provide you a direct link and the required forms

The novelty of SEM dataset is the non-posed reactions to autobiographical  and non-autobiographical visual stimulus data. The main contribution of SEM  database is the use of personalized images for each participant. These images  are photos of themselves or their relatives and friends both from the recent  and distant past.
The recorded data is provided in different data modalities: HD RGB, depth  and IR frames of the face, EEG signal and eye gaze data; which were recorded  using 4 different devices: a 30fps HD RGB camera, IR/Depth sensors (Kinect),  an eye tracker (Tobii eye tracker) and EEG sensors (Emotiv headset).
> Juan Manuel Fernandez Montenegro, Athanasios Gkelias, V. Argyriou, Emotion understanding using multimodal information based on autobiographical memories for Alzheimer’s patients, ACCVW, 2016
      • Spontaneous Emotion Multimodal Database (SEM-db)

>Games

  • Coming soon…