{"type":"ScholarlyArticle","title":"Multi-scale Spatial Representation Learning via Recursive Hermite Polynomial Networks","authors":[{"name":"Lin (Yuanbo) Wu","slug":"lin-yuanbo-wu","given":"Lin (Yuanbo)","family":"Wu","affiliation":"Key Laboratory of Knowledge Engineering with Big Data , Ministry of Education ; School of Computer Science and Information Engineering , Hefei University of Technology , Hefei 230009 , China."},{"name":"Deyin Liu","slug":"deyin-liu","given":"Deyin","family":"Liu","affiliation":"Anhui Provincial Key Laboratory of Multimodal Cognitive Computation , School of Artificial Intelligence , Anhui University , Hefei 230039 , China."},{"name":"Xiaojie Guo","slug":"xiaojie-guo","given":"Xiaojie","family":"Guo","affiliation":"Tianjin University , China"},{"name":"Richang Hong","slug":"richang-hong","given":"Richang","family":"Hong","affiliation":"Key Laboratory of Knowledge Engineering with Big Data , Ministry of Education ; School of Computer Science and Information Engineering , Hefei University of Technology , Hefei 230009 , China."},{"name":"Liangchen Liu","slug":"liangchen-liu","given":"Liangchen","family":"Liu","affiliation":"The University of Melbourne , Victoria 3052 , Australia."},{"name":"Rui Zhang","slug":"rui-zhang","given":"Rui","family":"Zhang","affiliation":""}],"abstract":"Multi-scale representation learning aims to leverage diverse features from different layers of Convolutional Neural Networks (CNNs) for boosting the feature robustness to scale variance. For dense prediction tasks, two key properties should be satisfied: the high spatial variance across convolutional layers, and the sub-scale granularity inside a convolutional layer for fine-grained features. To pursue the two properties, this paper proposes Recursive Hermite Polynomial Networks (RHP-Nets for short). The proposed RHP-Nets consist of two major components: 1) a dilated convolution to maintain the spatial resolution across layers, and 2) a family of Hermite polynomials over a subset of dilated grids, which recursively constructs sub-scale representations to avoid the artifacts caused by naively applying the dilation convolution. The resultant sub-scale granular features are fused via trainable Hermite coefficients to form the multi-resolution representations that can be fed into the next deeper layer, and thus allowing feature interchanging at all levels. Extensive experiments are conducted to demonstrate the efficacy of our design, and reveal its superiority over state-of-the-art alternatives on a variety of image recognition tasks. Besides, introspective studies are provided to further understand the properties of our method.","doi":"10.24963/ijcai.2022/204","url":"https://ijcai.org/proceedings/2022/204","pdf":"https://ijcai.org/proceedings/2022/0204.pdf","markdown":"https://ijcai.org/proceedings/2022/204.md","bibtex":"https://ijcai.org/proceedings/2022/bibtex/204","volume":{"slug":"2022","title":"Proceedings of the Thirty-First International Joint Conference on Artificial Intelligence","short_title":"IJCAI 2022","url":"https://ijcai.org/proceedings/2022/","year":2022,"location":"Vienna","dates":"23-29 July 2022","editors":"Luc De Raedt","isbn":"978-1-956792-00-3","publisher":"International Joint Conferences on Artificial Intelligence Organization","doi_prefix":"10.24963/ijcai.2022"},"section":"Computer Vision","track":"Main Track","number":204,"pages":{"first":1465,"last":1473},"keywords":["Computer Vision: Representation Learning"],"videos":[{"label":"Video #1","url":"https://slideslive.com/38986576","length_sec":105,"provider":"slideslive"},{"label":"Video #2","url":"https://slideslive.com/38984719","length_sec":315,"provider":"slideslive"}],"references":[{"position":1,"raw":"[Chen et al., 2018a] Kai Chen, Jiangmiao Pang, Jiaqi Wang, Yu Xiong, Xiaoxiao Li, Shuyang Sun, Wansen Feng, Ziwei Liu, Jianping Shi, Wanli Ouyang, Chen Change Loy, and Dahua Lin. mmdetection. In https://github.com/open-mmlab/mmdetection, pages -, 2018.","title":"mmdetection","authors":"Chen","year":"2018","venue":null,"doi":null,"url":"https://github.com/open-mmlab/mmdetection","internal_url":null},{"position":2,"raw":"[Chen et al., 2018b] L.-C. Chen, G. Papandreou, I. Kokkinos, K. Murphy, and A. L. Yuille. Deeplab: Semantic image segmen- tation with deep convolutional nets, atrous convolution, and fully connected crfs. IEEE Trans. Patt. Anal. Mach. Intel., 40(4):834- 848, April 2018.","title":"Deeplab: Semantic image segmentation with deep convolutional nets, atrous convolution, and fully connected crfs","authors":"Chen","year":"2018","venue":"IEEE Trans. Patt. Anal. Mach. Intel","doi":null,"url":null,"internal_url":null},{"position":3,"raw":"[Chiu et al., 2013] C. Chiu, M. A. Digman, and E. Gratton. Cell matrix remodeling ability shown by image spatial correlation. Journal of Biophysics, 2013(doi: 10.1155/2013/532030):1-8, July 2013.","title":"Cell matrix remodeling ability shown by image spatial correlation","authors":"C Chiu, M A Digman, E Gratton","year":"2013","venue":"Journal of Biophysics","doi":"10.1155/2013/532030):1-8","url":null,"internal_url":null},{"position":4,"raw":"[Choe and Shim, 2019] J. Choe and H. Shim. Attention-based dropout layer for weakly supervised object localization. In CVPR, pages 2219-2228, 2019.","title":"Attention-based dropout layer for weakly supervised object localization","authors":"Shim Choe, ] J Choe, H Shim","year":"2019","venue":"CVPR","doi":null,"url":null,"internal_url":null},{"position":5,"raw":"[Choe et al., 2020] J. Choe, S. J. Oh, S. Lee, S. Chun, Z. Akata, and H. Shim. Evaluating weakly supervised object localization methods right. In CVPR, pages 3133-3142, 2020.","title":"Evaluating weakly supervised object localization methods right","authors":"J Choe, S J Oh, S Lee, S Chun, Z Akata, H Shim","year":"2020","venue":"CVPR","doi":null,"url":null,"internal_url":null},{"position":6,"raw":"[Cordts et al., 2016] M. Cordts, M. Omran, S. Ramos, T. Rehfeld, M. Enzweiler, R. Benenson, U. Franke, S. Roth, and B. Schiele. The cityscapes dataset for semantic urban scene understanding. In CVPR, 2016.","title":"The cityscapes dataset for semantic urban scene understanding","authors":"M Cordts, M Omran, S Ramos, T Rehfeld, M Enzweiler, R Benenson, U Franke, S Roth, B Schiele","year":"2016","venue":"CVPR","doi":null,"url":null,"internal_url":null},{"position":7,"raw":"[Fan et al., 2020] B. Fan, L. Wang, R. Zhang, Z. Guo, Y. Zhao, R. Li, and W. Gong. Contextual multi-scale feature learning for person re-identification. In ACM Multimedia, 2020.","title":"Contextual multi-scale feature learning for person re-identification","authors":"B Fan, L Wang, R Zhang, Z Guo, Y Zhao, R Li, W Gong","year":"2020","venue":"ACM Multimedia","doi":null,"url":null,"internal_url":null},{"position":8,"raw":"[Fuchs et al., 2019] A. Fuchs, R. Priewald, and F. Pernkopf. Recur- rent dilated densenets for a time-series segmentation task. In In- ternational Conference on Machine Learning and Applications, pages 75-80, 2019.","title":"Recurrent dilated densenets for a time-series segmentation task","authors":"A Fuchs, R Priewald, F Pernkopf","year":"2019","venue":"International Conference on Machine Learning and Applications","doi":null,"url":null,"internal_url":null},{"position":9,"raw":"[Ge et al., 2018] R. Ge, J. D. Lee, and T. Ma. Learning one-hidden- layer neural networks with landscape design. In ICLR, pages -, 2018.","title":"Learning one-hiddenlayer neural networks with landscape design","authors":"R Ge, J D Lee, T Ma","year":"2018","venue":"ICLR","doi":null,"url":null,"internal_url":null},{"position":10,"raw":"[Glorot et al., 2011] X. Glorot, A. Bordes, and Y. Bengio. Deep sparse rectifier neural networks. In International conference on artificial intelligence and statistics, pages 315-323, 2011.","title":"Deep sparse rectifier neural networks","authors":"X Glorot, A Bordes, Y Bengio","year":"2011","venue":"International conference on artificial intelligence and statistics","doi":null,"url":null,"internal_url":null},{"position":11,"raw":"[Gong and Poellabauer, 2018] Y. Gong and C. Poellabauer. Impact of aliasing on deep cnn-based end-to-end acoustic models. In International Speech, 2018.","title":"Impact of aliasing on deep cnn-based end-to-end acoustic models","authors":"Y Gong, C Poellabauer","year":"2018","venue":"International Speech","doi":null,"url":null,"internal_url":null},{"position":12,"raw":"[He et al., 2016] K. He, X. Zhang, S. Ren, and J. Sun. Deep resid- ual learning for image recognition. In CVPR, 2016.","title":"Deep residual learning for image recognition","authors":"K He, X Zhang, S Ren, J Sun","year":"2016","venue":"CVPR","doi":null,"url":null,"internal_url":null},{"position":13,"raw":"[He et al., 2017] Kaiming He, Georgia Gkioxari, Piotr Dollár, and Ross Girshick. Mask r-cnn. In ICCV, pages -, 2017.","title":"Kaiming He, Georgia Gkioxari","authors":"He","year":"2017","venue":"ICCV","doi":null,"url":null,"internal_url":null},{"position":14,"raw":"[Huang et al., 2017] G. Huang, Z. Liu, L. van der Maaten, and K. Q. Weinberger. Densely connected convolutional networks. In CVPR, pages 4700-4708, 2017.","title":"Densely connected convolutional networks","authors":"G Huang, Z Liu, L Van Der Maaten, K Q Weinberger","year":"2017","venue":"CVPR","doi":null,"url":null,"internal_url":null},{"position":15,"raw":"[Lan et al., 2018] X. Lan, X. Zhu, and S. Gong. Person search by multi-scale matching. In ECCV, pages -, 2018.","title":"Person search by multi-scale matching","authors":"Lan","year":"2018","venue":"ECCV","doi":null,"url":null,"internal_url":null},{"position":16,"raw":"[Li et al., 2018] Y. Li, X. Zhang, and D. Chen. Csrnet: Dilated convolutional neural networks for understanding the highly con- gested scenes. In CVPR, 2018.","title":"Csrnet: Dilated convolutional neural networks for understanding the highly congested scenes","authors":"Y Li, X Zhang, D Chen","year":"2018","venue":"CVPR","doi":null,"url":null,"internal_url":null},{"position":17,"raw":"[Li et al., 2019] Y. Li, Z. Kuang, Y. Chen, and W. Zhang. Data- driven neuron allocation for scale aggregation networks. In CVPR, pages 11526-11534, 2019.","title":"Datadriven neuron allocation for scale aggregation networks","authors":"Y Li, Z Kuang, Y Chen, W Zhang","year":"2019","venue":"CVPR","doi":null,"url":null,"internal_url":null},{"position":18,"raw":"[Lin et al., 2014] Tsung-Yi Lin, Michael Maire, Serge Belongie, James Hays, Pietro Perona, Deva Ramanan, Piotr Dollar, and Larry Zitnick. Microsoft coco: Common objects in context. In ECCV, pages -, 2014.","title":"Microsoft coco: Common objects in context","authors":"Tsung-Yi Lin, Michael Maire, Serge Belongie, James Hays, Pietro Perona, Deva Ramanan, Piotr Dollar, Larry Zitnick","year":"2014","venue":"ECCV","doi":null,"url":null,"internal_url":null},{"position":19,"raw":"[Lin et al., 2017a] T.-Y. Lin, P. Dollár, R. Girshick, K. He, B. Har- iharan, and S. Belongie. Feature pyramid networks for object detection. In CVPR, pages 2117-2125, 2017.","title":"Feature pyramid networks for object detection","authors":"Lin","year":"2017","venue":"CVPR","doi":null,"url":null,"internal_url":null},{"position":20,"raw":"[Lin et al., 2017b] Tsung-Yi Lin, Priya Goyal, Ross Girshick, Kaiming He, and Piotr Dollár. Focal loss for dense object de- tection. In ICCV, pages -, 2017.","title":"Focal loss for dense object detection","authors":"Lin","year":"2017","venue":"ICCV","doi":null,"url":null,"internal_url":null},{"position":21,"raw":"[Liu et al., 2010] T. Liu, Z. Yuan, J. Sun, J. Wang, N. Zheng, X. Tang, and H.-Y. Shum. Learning to detect a salient object. IEEE Trans. Pat. Anal. Mach. Intel., 33(2):353-367, April 2010.","title":"Learning to detect a salient object","authors":"T Liu, Z Yuan, J Sun, J Wang, N Zheng, X Tang, H.-Y Shum","year":"2010","venue":"IEEE Trans. Pat. Anal. Mach. Intel","doi":null,"url":null,"internal_url":null},{"position":22,"raw":"[Liu et al., 2017] H. Liu, J. Feng, Z. Jie, K. Jayashree, B. Zhao, M. Qi, J. Jiang, and S. Yan. Neural person search machines. In ICCV, pages 493-501, 2017.","title":"Neural person search machines","authors":"H Liu, J Feng, Z Jie, K Jayashree, B Zhao, M Qi, J Jiang, S Yan","year":"2017","venue":"ICCV","doi":null,"url":null,"internal_url":null},{"position":23,"raw":"[Lokhande et al., 2020] V. Suresh Lokhande, S. Tasneeyapant, A. Venkatesh, S. N. Ravi, and V. Singh. Generating accurate pseudo-labels in semi-supervised learning and avoiding overcon- fident predictions via hermite polynomial activations. In CVPR, pages 11435-11443, 2020.","title":"Generating accurate pseudo-labels in semi-supervised learning and avoiding overconfident predictions via hermite polynomial activations","authors":"V Suresh Lokhande, S Tasneeyapant, A Venkatesh, S N Ravi, V Singh","year":"2020","venue":"CVPR","doi":null,"url":null,"internal_url":null},{"position":24,"raw":"[Long et al., 2015] J. Long, E. Shelhamer, and T. Darrell. Fully convolutional networks for semantic segmentation. In CVPR, 2015.","title":"Fully convolutional networks for semantic segmentation","authors":"J Long, E Shelhamer, T Darrell","year":"2015","venue":"CVPR","doi":null,"url":null,"internal_url":null},{"position":25,"raw":"[Lu et al., 2019] Xin Lu, Buyu Li, Yuxin Yue, Quanquan Li, and Junjie Yan. Grid r-cnn. In CVPR, pages -, 2019.","title":"Grid r-cnn","authors":"Xin Lu, Buyu Li, Yuxin Yue, Quanquan Li, Junjie Yan","year":"2019","venue":"CVPR","doi":null,"url":null,"internal_url":null},{"position":26,"raw":"[Martens, 2006] J.-B. Martens. The hermite transform: A survey. EURASIP Journal on Advances in Signal Processing, 2006:1-20, 2006.","title":"The hermite transform: A survey","authors":"J.-B Martens","year":"2006","venue":"EURASIP Journal on Advances in Signal Processing","doi":null,"url":null,"internal_url":null},{"position":27,"raw":"[Newell et al., 2016] A. Newell, K. Yang, and J. Deng. Stacked hourglass network for human pose estimation. In ECCV, 2016.","title":"Stacked hourglass network for human pose estimation","authors":"A Newell, K Yang, J Deng","year":"2016","venue":"ECCV","doi":null,"url":null,"internal_url":null},{"position":28,"raw":"[Pang et al., 2019] Jiangmiao Pang, Kai Chen, Jianping Shi, Hua- jun Feng, Wanli Ouyang, and Dahua Lin. Libra r-cnn: Towards balanced learning for object detection. In CVPR, pages -, 2019.","title":"Libra r-cnn: Towards balanced learning for object detection","authors":"Pang","year":"2019","venue":"CVPR","doi":null,"url":null,"internal_url":null},{"position":29,"raw":"[Pauwels et al., 1995] E. J. Pauwels, L. J. Van Gool, P. Fiddelaers, and T. Moons. An extended class of scale-invariant and recur- sive scale space filters. IEEE Trans. Patt. Anal. Mach. Intel., 17(4):691-701, 1995.","title":"An extended class of scale-invariant and recursive scale space filters","authors":"E J Pauwels, L J Van Gool, P Fiddelaers, T Moons","year":"1995","venue":"IEEE Trans. Patt. Anal. Mach. Intel","doi":null,"url":null,"internal_url":null},{"position":30,"raw":"[Qian et al., 2017] X. Qian, Y. Fu, Y. Jiang, T. Xiang, and X. Xue. Multi-scale deep learning architectures for person re- identification. pages 5399-5408, 2017.","title":"Multi-scale deep learning architectures for person reidentification","authors":"X Qian, Y Fu, Y Jiang, T Xiang, X Xue","year":"2017","venue":null,"doi":null,"url":null,"internal_url":null},{"position":31,"raw":"[Rasiah et al., 1997] A.I. Rasiah, R. Togneri, and Y. Attikiouzel. Modelling 1-d signals using hermite basis functions. IEEE proceedings-Vision, Image and Signal Processing, 144(6):345- 354, 1997.","title":"Modelling 1-d signals using hermite basis functions","authors":"A I Rasiah, R Togneri, Y Attikiouzel","year":"1997","venue":"IEEE proceedings-Vision, Image and Signal Processing","doi":null,"url":null,"internal_url":null},{"position":32,"raw":"[Ren et al., 2015] S. Ren, K. He, R. Girshick, and J. Sun. Faster r-cnn: Towards real-time object detection with region proposal networks. In NIPS, pages -, 2015.","title":"Faster r-cnn: Towards real-time object detection with region proposal networks","authors":"S Ren, K He, R Girshick, J Sun","year":"2015","venue":"NIPS","doi":null,"url":null,"internal_url":null},{"position":33,"raw":"[Schuster et al., 2019] R. Schuster, O. Wasenmuller, C. Unger, and D. Stricker. Sdc-stacked dilated convolution: A unified descrip- tor network for dense matching tasks. In CVPR, pages 2556- 2565, 2019.","title":"Sdc-stacked dilated convolution: A unified descriptor network for dense matching tasks","authors":"R Schuster, O Wasenmuller, C Unger, D Stricker","year":"2019","venue":"CVPR","doi":null,"url":null,"internal_url":null},{"position":34,"raw":"[Shen et al., 2019] Y. Shen, R. Ji, X. Hong, F. Zheng, X. Guo, Y. Wu, and F. Huang. A part power set model for scale-free per- son retrieval. In IJCAI, pages 3397-3403, 2019.","title":"A part power set model for scale-free person retrieval","authors":"Y Shen, R Ji, X Hong, F Zheng, X Guo, Y Wu, F Huang","year":"2019","venue":"IJCAI","doi":null,"url":null,"internal_url":"/proceedings/2019/471"},{"position":35,"raw":"[Singh and Lee, 2017] K. K. Singh and Y. J. Lee. Hide-and-seek: Forcing a network to be meticulous for weakly-supervised object and action localization. In ICCV, 2017.","title":"Hide-and-seek: Forcing a network to be meticulous for weakly-supervised object and action localization","authors":"K K Singh, Y J Lee","year":"2017","venue":"ICCV","doi":null,"url":null,"internal_url":null},{"position":36,"raw":"[Sun et al., 2019] K. Sun, B. Xiao, D. Liu, and J. Wang. Deep high- resolution representation learning for human pose estimation. In CVPR, 2019.","title":"Deep highresolution representation learning for human pose estimation","authors":"K Sun, B Xiao, D Liu, J Wang","year":"2019","venue":"CVPR","doi":null,"url":null,"internal_url":null},{"position":37,"raw":"[Szegedy et al., 2016] C. Szegedy, S. Ioffe, V. Vanhouck, and A. A. Alemi. Inception-4v, inception-resnet and the impact of residual connections on learning. In AAAI, 2016.","title":"Inception-4v, inception-resnet and the impact of residual connections on learning","authors":"C Szegedy, S Ioffe, V Vanhouck, A A Alemi","year":"2016","venue":"AAAI","doi":null,"url":null,"internal_url":null},{"position":38,"raw":"[Takahashi and Mitsufuji, 2021] N. Takahashi and Y. Mitsufuji. Densely connected multidilated convolutional networks for dense prediction tasks. In CVPR, pages 993-1002, 2021.","title":"Densely connected multidilated convolutional networks for dense prediction tasks","authors":"N Takahashi, Y Mitsufuji","year":"2021","venue":"CVPR","doi":null,"url":null,"internal_url":null},{"position":39,"raw":"[Wah et al., 2010] C. Wah, S. Branson, P. Welinder, P. Perona, and S. Belongie. Caltech-ucsd birds 200. In Technical report CNS- TR-2010-001, California Insititute of TEchnology, 2010.","title":"Caltech-ucsd birds 200","authors":"C Wah, S Branson, P Welinder, P Perona, S Belongie","year":"2010","venue":null,"doi":null,"url":null,"internal_url":null},{"position":40,"raw":"[Wang et al., 2018] P. Wang, P. Chen, Y. Yuan, D. Liu, Z. Huang, X. Hou, and G. Cottrell. Understanding convolution for semantic segmentation. In WACV, 2018.","title":"Understanding convolution for semantic segmentation","authors":"P Wang, P Chen, Y Yuan, D Liu, Z Huang, X Hou, G Cottrell","year":"2018","venue":"WACV","doi":null,"url":null,"internal_url":null},{"position":41,"raw":"[Wang et al., 2020] X. Wang, S. Zhang, Z. Yu, L. Feng, and W. Zhang. Scale-equalizing pyramid convolutions for object de- tection. In CVPR, pages 13359-13368, 2020.","title":"Scale-equalizing pyramid convolutions for object detection","authors":"Wang","year":"2020","venue":"CVPR","doi":null,"url":null,"internal_url":null},{"position":42,"raw":"[Xiao et al., 2017] T. Xiao, S. Li, B. Wang, L. Lin, and X. Wang. Joint detection and identification feature learning for person search. In CVPR, pages 3376-3385, 2017.","title":"Joint detection and identification feature learning for person search","authors":"T Xiao, S Li, B Wang, L Lin, X Wang","year":"2017","venue":"CVPR","doi":null,"url":null,"internal_url":null},{"position":43,"raw":"[Yu et al., 2017] F. Yu, V. Koltun, and T. Funkhouser. Dilated resid- ual networks. In CVPR, pages 472-480, 2017.","title":"Dilated residual networks","authors":"F Yu, V Koltun, T Funkhouser","year":"2017","venue":"CVPR","doi":null,"url":null,"internal_url":null},{"position":44,"raw":"[Yun et al., 2019] S. Yun, D. Han, S. J. Oh, S. Chun, J. Choe, and Y. Yoo. Cutmix: regularization strategy to train strong classifiers with localizable features. In ICCV, 2019.","title":"Cutmix: regularization strategy to train strong classifiers with localizable features","authors":"S Yun, D Han, S J Oh, S Chun, J Choe, Y Yoo","year":"2019","venue":"ICCV","doi":null,"url":null,"internal_url":null},{"position":45,"raw":"[Zhang et al., 2018a] X. Zhang, Y. Wei, J. Feng, Y. Yang, and T. Huang. Adversarial complementary learning for weakly su- pervised object localization. In CVPR, pages 1325-1334, 2018.","title":"Adversarial complementary learning for weakly supervised object localization","authors":"Zhang","year":"2018","venue":"CVPR","doi":null,"url":null,"internal_url":null},{"position":46,"raw":"[Zhang et al., 2018b] X. Zhang, Y. Wei, G. Kang, Y. Yang, and T. Huang. Self-produced guidance for weakly-supervised object localization. In ECCV, 2018.","title":"Self-produced guidance for weakly-supervised object localization","authors":"Zhang","year":"2018","venue":"ECCV","doi":null,"url":null,"internal_url":null},{"position":47,"raw":"[Zheng et al., 2017] L. Zheng, H. Zhang, S. Sun, M. Chandraker, Y. Yang, and Q. Tian. Person re-identification in the wild. In CVPR, pages 1367-1376, 2017.","title":"Person re-identification in the wild","authors":"L Zheng, H Zhang, S Sun, M Chandraker, Y Yang, Q Tian","year":"2017","venue":"CVPR","doi":null,"url":null,"internal_url":null},{"position":48,"raw":"[Zhou et al., 2016] B. Zhou, A. Khosla, A. Lapedriza, A. Oliva, and A. Torralba. Learning deep features for discriminative localiza- tion. In CVPR, pages 2921-2929, 2016.","title":"Learning deep features for discriminative localization","authors":"B Zhou, A Khosla, A Lapedriza, A Oliva, A Torralba","year":"2016","venue":"CVPR","doi":null,"url":null,"internal_url":null}],"license_note":"Copyright International Joint Conferences on Artificial Intelligence."}