A Preliminary Study on Integrating Camera and LiDAR Pedestrian Detections Adaptive to Surrounding Environmental Conditions

Results of pedestrian detectors from in-vehicle sensors still have room for improvement in real environments. Therefore, we have proposed estimation systems on the reliability (e.g., probability of oversight and misdetection) of pedestrian detection adaptive to surrounding environmental conditions. We expect to obtain integrated detection results that compensate for the strong and weak points of each sensor by referring to such reliabilities. This report presents a preliminary study on the construction method of the integrated detector referring to the reliabilities adaptive to surrounding environmental conditions.

[1]  Bill Triggs,et al.  Histograms of oriented gradients for human detection , 2005, 2005 IEEE Computer Society Conference on Computer Vision and Pattern Recognition (CVPR'05).

[2]  Hiroshi Murase,et al.  Estimating the Scene-wise Reliability of LiDAR Pedestrian Detectors , 2018, 2018 21st International Conference on Intelligent Transportation Systems (ITSC).

[3]  Yasuhiro Suzuki,et al.  Pedestrian detection and tracking using in-vehicle lidar for automotive application , 2011, 2011 IEEE Intelligent Vehicles Symposium (IV).

[4]  Hiroshi Murase,et al.  Pedestrian detection from sparse point-cloud using 3DCNN , 2018, 2018 International Workshop on Advanced Image Technology (IWAIT).

[5]  Pietro Perona,et al.  Fast Feature Pyramids for Object Detection , 2014, IEEE Transactions on Pattern Analysis and Machine Intelligence.

[6]  Hiroshi Murase,et al.  Efficient pedestrian scanning by active scan LIDAR , 2018, 2018 International Workshop on Advanced Image Technology (IWAIT).

[7]  Dushyant Rao,et al.  Vote3Deep: Fast object detection in 3D point clouds using efficient convolutional neural networks , 2016, 2017 IEEE International Conference on Robotics and Automation (ICRA).