Applying Faster R-CNN in Extremely Low-Resolution Thermal Images for People Detection

In today's cities, it is increasingly normal to see different systems based on Artificial Intelligence (AI) that help citizens and government institutions in their daily lives. This is possible thanks to the Internet of Things (IoT). In this paper we present a solution using low-resolution thermal sensors in combination of deep learning to detect people in the images generated by those sensors. To verify whether the deep learning techniques are appropriate for this type of images of such low resolution, we have implement a Faster Region-Convolutional Neural Network. The results obtained are hopeful and undoubtedly encourage to continue improving this research line. With a perception of 72.85% and given the complexity of the problem presented we consider the results obtained to be highly satisfactory and it encourages us to continue improving the work presented in this article.

[1]  Kai Oliver Arras,et al.  People detection in RGB-D data , 2011, 2011 IEEE/RSJ International Conference on Intelligent Robots and Systems.

[2]  Kaiming He,et al.  Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks , 2015, IEEE Transactions on Pattern Analysis and Machine Intelligence.

[3]  Tsong-Yi Chen,et al.  A People Counting System Based on Face-Detection , 2010, 2010 Fourth International Conference on Genetic and Evolutionary Computing.

[4]  Pei Li,et al.  People counting based on head detection combining Adaboost and CNN in crowded surveillance environment , 2016, Neurocomputing.

[5]  Rachel Cardell-Oliver,et al.  Occupancy Estimation Using a Low-Pixel Count Thermal Imager , 2016, IEEE Sensors Journal.

[6]  Alberto Cerpa,et al.  ThermoSense: Occupancy Thermal Based Sensing for HVAC Control , 2013, BuildSys@SenSys.

[7]  Yunhao Liu,et al.  Non-Invasive Detection of Moving and Stationary Human With WiFi , 2015, IEEE Journal on Selected Areas in Communications.

[8]  Suk Lee,et al.  Development of PIR sensor based indoor location detection system for smart home , 2006, 2006 SICE-ICASE International Joint Conference.

[9]  Yutaka Hata,et al.  Multi-human Locating in Real Environment by Thermal Sensor , 2013, 2013 IEEE International Conference on Systems, Man, and Cybernetics.

[10]  Fabio Salice,et al.  Indoor Human Detection Based on Thermal Array Sensor Data and Adaptive Background Estimation , 2017, Journal of Computer and Communications.

[11]  Yongwoo Jeong,et al.  Probabilistic method to determine human subjects for low-resolution thermal imaging sensor , 2014, 2014 IEEE Sensors Applications Symposium (SAS).

[12]  Huchuan Lu,et al.  A real-time effective system for tracking passing people using a single camera , 2008, 2008 7th World Congress on Intelligent Control and Automation.

[13]  Alberto Cerpa,et al.  ThermoSense: thermal array sensor networks in building management , 2013, SenSys '13.

[14]  Sasha Colby,et al.  Projections of the Size and Composition of the U.S. Population: 2014 to 2060. Population Estimates and Projections. Current Population Reports. P25-1143. , 2015 .