====== Image Understanding ====== |< 100% 33% 66%>| ^Dauer^4 SWS^ ^Art^Seminar^ ^Semester^SS2016^ ^Vortragende^Prof. Michael Beetz, Ferenc Balint-Benczedi, Feroz Ahmed Siddiky, Thiemo Wiedemeyer, Jan-Hendrik Worch^ ^Sprache^Deutsch, Englisch^ ^Termine^ Mo., 10:00 - 12:00, Ort: TAB 1.58^ ^Bemerkungen^Vorlesungsbeginn: 18.04.2016^ \\ \\ Organizational Issues and Materials can be found at our [[https://elearning.uni-bremen.de/dispatch.php/course/overview?cid=42f287094015f244f6b4c3376a847141|Stud.IP page]] ===== Description ===== The seminar will deal with the challenges of semantic perception in the context of robotics, presenting various aspects of it. Students will be presented with an overview of the field followed by individual presentations and reports of pre-defined topics. ===== Literature ===== ==== Segmentation ==== Weakly supervised graph based semantic segmentation by learning communities of image-parts http://www.cv-foundation.org/openaccess/content_iccv_2015/papers/Pourian_Weakly_Supervised_Graph_ICCV_2015_paper.pdf Decision Making under Uncertain Segmentations http://ieeexplore.ieee.org/xpls/abs_all.jsp?arnumber=7139359 ==== Features ==== KAZE Features **+** Fast Explicit Diffusion for Accelerated Features in Nonlinear Scale Spaces http://www.robesafe.com/personal/pablo.alcantarilla/papers/Alcantarilla12eccv.pdf http://www.robesafe.com/personal/pablo.alcantarilla/papers/Alcantarilla13bmvc.pdf B-SHOT: A Binary Feature Descriptor for Fast and Efficient Keypoint Matching on 3D Point Clouds http://ieeexplore.ieee.org/stamp/stamp.jsp?arnumber=7353630 Rotation and Translation Invariant 3D Descriptor for Surfaces http://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=7353450 ==== Object Detection, Recognition and Tracking ==== Real-time Pose Detection and Tracking of Hundreds of Objects **+** SimTrack: A Simulation-based Framework for Scalable Real-time Object Pose Detection and Tracking http://www.karlpauwels.com/downloads/tcsvt_2015/Pauwels_IEEE_TCSVT_2015.pdf http://www.karlpauwels.com/downloads/iros_2015/Pauwels_IROS_2015.pdf Surface Oriented Traverse for Robust Instance Detection in RGB-D http://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=7353983 RGB-D Object Modelling for Object Recognition and Tracking http://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=7353360 Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks http://arxiv.org/abs/1506.01497 Rich feature hierarchies for accurate object detection and semantic segmentation http://arxiv.org/abs/1311.2524 Efficient RGB-D Object Categorization Using cascaded Ensembles of Randomized Decision Trees http://ieeexplore.ieee.org/stamp/stamp.jsp?arnumber=7139358 Robust 3D tracking of Unknown Objects http://ieeexplore.ieee.org/xpl/articleDetails.jsp?tp=&arnumber=7139520 Depth-Based Tracking with Physical Constraints for Robot Manipulation http://homes.cs.washington.edu/~tws10/DepthBasedTracking.pdf ==== Affordances ==== AfNet: The Affordance Network http://link.springer.com/chapter/10.1007%2F978-3-642-37331-2_39 Affordance detection of Tool parts from Geometric Features http://www.visionmeetscognition.org/fpic2014/Camera_Ready/Paper%2035.pdf Long-term human affordance maps http://dx.doi.org/10.1109/IROS.2015.7354193 ==== Deep Learning ==== Visualizing and Understanding Convolutional Networks https://www.cs.nyu.edu/~fergus/papers/zeilerECCV2014.pdf DeepFace: Closing the Gap to Human-Level Performance in Face Verification https://www.cs.toronto.edu/~ranzato/publications/taigman_cvpr14.pdf MoDeep: A Deep Learning Framework Using Motion Features for Human Pose Estimation http://cs.nyu.edu/~ajain/accv2014/paper.pdf Joint Training of a Convolutional Network and a Graphical Model for Human Pose Estimation https://www.robots.ox.ac.uk/~vgg/rg/papers/tompson2014.pdf Flowing ConvNets for Human Pose Estimation in Videos https://www.robots.ox.ac.uk/~vgg/publications/2015/Pfister15a/pfister15a.pdf Multimodal deep learning for robust RGB-D object recognition http://arxiv.org/pdf/1507.06821v2.pdf RGB-D Object Recognition and Pose Estimation Based on Pre-Trained Convolutional Neural Network Features https://www.ais.uni-bonn.de/papers/ICRA_2015_Schwarz_RGB-D-Objects_Transfer-Learning.pdf ==== Unsupervised Deep Learning ==== Unsupervised Learning of Invariant Feature Hierarchies with Applications to Object Recognition http://yann.lecun.com/exdb/publis/pdf/ranzato-cvpr-07.pdf Convolution Deep Belief Networks for Scalable Unsupervised Learning of Hierarchical Representations http://web.eecs.umich.edu/~honglak/icml09-ConvolutionalDeepBeliefNetworks.pdf Sparse Feature Learning for Deep Belief Networks https://papers.nips.cc/paper/3363-sparse-feature-learning-for-deep-belief-networks.pdf Efficient sparse coding algorithms https://papers.nips.cc/paper/2979-efficient-sparse-coding-algorithms.pdf ==== Human Detection and Tracking ==== Automatic initialization for skeleton tracking in optical motion capture http://dx.doi.org/10.1109/ICRA.2015.7139260 Unsupervised robot learning to predict person motion http://dx.doi.org/10.1109/ICRA.2015.7139254 Pose estimation for a partially observable human body from RGB-D cameras http://dx.doi.org/10.1109/IROS.2015.7354068 Real-time full-body human attribute classification in RGB-D using a tessellation boosting approach http://dx.doi.org/10.1109/IROS.2015.7353541 ==== Action Recognition ==== Learning symbolic representations of actions from human demonstrations http://dx.doi.org/10.1109/ICRA.2015.7139728 Fast Target Prediction of Human Reaching Motion for Cooperative Human-Robot Manipulation Tasks Using Time Series Classification http://dx.doi.org/10.1109/ICRA.2015.7140066 Effective 3D action recognition using EigenJoints http://dx.doi.org/10.1016/j.jvcir.2013.03.001 Sequence of the most informative joints (SMIJ): A new representation for human skeletal action recognition http://dx.doi.org/10.1016/j.jvcir.2013.04.007 Unsupervised Temporal Segmentation of Repetitive Human Actions Based on Kinematic Modeling and Frequency Analysis http://arxiv.org/abs/1512.04115 sEMG-based decoding of detailed human intentions from finger-level hand motions http://dx.doi.org/10.1109/IROS.2015.7353982 Human motion classification and recognition using wholebody contact force http://dx.doi.org/10.1109/IROS.2015.7353979 Context-based intent understanding using an Activation Spreading architecture http://dx.doi.org/10.1109/IROS.2015.7353791 A framework for unsupervised online human reaching motion recognition and early prediction http://dx.doi.org/10.1109/IROS.2015.7353706 Human intention inference and motion modeling using approximate E-M with online learning http://dx.doi.org/10.1109/IROS.2015.7353614 ==== RoboSherlock ==== **These four papers count as one block, i.e. they have to be presented together.** RoboSherlock: Unstructured Information Processing for Robot Perception http://ai.uni-bremen.de/_media/paper/beetz15robosherlock.pdf RoboSherlock: Unstructured Information Processing Framework for Robotic Perception http://dx.doi.org/10.1007/978-3-319-26327-4_8 Pervasive 'Calm' Perception for Autonomous Robotic Agents http://ai.uni-bremen.de/_media/paper/Wiedemeyer15pervasive.pdf Perception for Everyday Human Robot Interaction http://dx.doi.org/10.1007/s13218-015-0400-1