Section: New Results

Introduction

This year Stars has proposed new results related to its three main research axes : perception for activity recognition, semantic activity recognition and software engineering for activity recognition.

Perception for Activity Recognition

Participants : François Brémond, Juan Diego Gonzales Zuniga, Abhijit Das, Antitza Dancheva, Furqan Muhammad Khan, Michal Koperski, Thi Lan Anh Nguyen, Remi Trichet, Ujjwal Ujjval, Srijan Das, Vikas Thamizharasan, Monique Thonnat.

The new results for perception for activity recognition are:

• Late Fusion of multiple convolutional layers for pedestrian detection (see 7.2)

• Deep Learning applied on Embedded Systems for People Tracking (see 7.3)

• Cross Domain Residual Transfer Learning for Person Re-identification (see 7.4)

• Face-based Attribute Classification (see 7.5)

• Face Attribute manipulation

• From attribute-labels to faces: face generation using a conditional generative adversarial network (see 7.6)

• Face analysis in structured light images (see 7.7)

Semantic Activity Recognition

Participants : François Brémond, Antitza Dantcheva, Farhood Negin, Thanh Hung Nguyen, Michal Koperski, Srijan Das, Kaustubh Sakhalkar, Arpit Chaudhary, Abhishek Goel, Abdelrahman Abubakr, Abhijit Das, Yaohui Wang, S L Happy, Alexandra König, Guillaume Sacco, Philippe Robert, Soumik Mallick, Julien Badie, Monique Thonnat.

For this research axis, the contributions are :

• Deep-Temporal LSTM for Daily Living Action Recognition (see 7.8)

• A New Hybrid Architecture for Human Activity Recognition from RGB-D videos (see 7.10)

• Where to focus on for Human Action Recognition? (see 7.11)

• Online temporal detection of daily-living human activities in long untrimmed video streams (see 7.12)

• Activity Detection in Long-term Untrimmed Videos (see 7.13)

• Video based face analysis for health monitoring (see 7.14)

• Mobile biometrics (see 7.15)

• Comparing methods for assessment of facial dynamics in patients with major neurocognitive disorders (see 7.16)

• Combating the issue of low sample size in facial expression recognition (see 7.17)

• Serious exergames for Cognitive Stimulation (see 7.18)

• Fully Automatic Speech-Based Analysis of the Semantic Verbal Fluency Task (see 7.19)

• Language Modelling in the Clinical Semantic Verbal Fluency Task (see 7.19.2)

• Telephone-based Dementia Screening I: Automated Semantic Verbal Fluency Assessment (see 7.19.3)

• Automatic Detection of Apathy using Acoustic Markers extracted from Free Emotional Speech and using Automatic Speech Analysis (see 7.19.4)

• Monitoring the Behaviors of Retail Customers (see 7.20)

Software Engineering for Activity Recognition

Participants : Sabine Moisan, Annie Ressouche, Jean-Paul Rigault, Ines Sarray, Daniel Gaffé, Julien Badie, François Brémond, Minh Khue Phan Tran.

The contributions for this research axis are:

• A Synchronous Approach to Activity Recognition (see 7.21)

• A Probabilistic Activity Description Language (see 7.22)