{"title":"A New Image Watermarking Scheme for Efficient Tamper Detection, Localization and Recovery","authors":"Faranak Tohidi, M. Paul","doi":"10.1109/ICMEW.2019.0-117","DOIUrl":"https://doi.org/10.1109/ICMEW.2019.0-117","url":null,"abstract":"Due to the powerful image editing tools, image integrity protection is becoming important, for instance, in a court of law, news reports, insurance claims or telemedicine. Digital image watermarking is a technique to detect, localize and recover tampered image but the accuracy of performance by the existing methods is still an issue. This paper proposes a new watermarking embedding mechanism for effective image authentication with a recovery scheme. In the method, a host image is divided into blocks and the feature of each block is embedded in the block itself and another block as watermark. For detecting tampering, the extracted watermarked data is checked with source and destination block leading to more accuracy. The recovery data is extracted from the destination block which is intact. The proposed scheme achieves better results in terms of tamper detection accuracy and the quality of recovered image compared with the state-of-art schemes.","PeriodicalId":302235,"journal":{"name":"2019 IEEE International Conference on Multimedia & Expo Workshops (ICMEW)","volume":"86 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2019-08-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124885739","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"Single Image Haze Removal by Feature Mapping","authors":"Yu Zhou, Feiniu Yuan, Xue Xia, Ya Li","doi":"10.1109/ICMEW.2019.00032","DOIUrl":"https://doi.org/10.1109/ICMEW.2019.00032","url":null,"abstract":"To restore a clear image from a single hazy image, many priors have been proposed for transmission estimation, but these priors often fail in some cases. In this paper, we propose a regression mapping model based on hazy features for haze removal. We fisrt propose sixteen hazy features for mapping, which include the dark channel prior, hazy measures and statistical features. Support Vector Regression (SVR) is adopted to learn the mapping between these hazy features and corresponding transmissions. Then we randomly select patches of synthetic haze and corresponding transmissions from a synthetic dataset to train the mapping model. Experimental results on both natural and synthetic images show that our regression based method achieves significantly better performance than existing state-of-the-art methods.","PeriodicalId":302235,"journal":{"name":"2019 IEEE International Conference on Multimedia & Expo Workshops (ICMEW)","volume":"93 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2019-07-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125112600","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"Annotating 3D Models and Their Parts via Deep Feature Embedding","authors":"K. Omata, T. Furuya, Ryutarou Ohbuchi","doi":"10.1109/ICMEW.2019.00090","DOIUrl":"https://doi.org/10.1109/ICMEW.2019.00090","url":null,"abstract":"Need to organize 3D shape data has prompted studies on comparison and retrieval of 3D shape models. Being able to query 3D shape models by words, in addition to 3D model examples and 2D sketches, would be quite beneficial. This paper proposes a method to associate whole 3D models (e.g., automobile) as well as their parts (e.g., tire, body, engine) with word labels so that the 3D model can be queried by words. The associations between 3D shapes and words are learned from a dataset of 3D models whose whole model and segmented parts are labeled with words. Feature vectors of these words (distributed representation) and feature vectors of whole and partial geometries of 3D models are embedded, by Word Shape embedding Network (WSN) into a common feature embedding space. As the word feature vectors are learned by Word2Vec trained on Wikipedia corpus, the common embedding space can be queried by a wide variety of words that are not included in the labeled 3D model dataset. Experimental evaluation has shown that, with the proposed algorithm, 3D shape can be queried by labels of either whole or part shape, or labels that are semantically close but not included in the original 3D model dataset.","PeriodicalId":302235,"journal":{"name":"2019 IEEE International Conference on Multimedia & Expo Workshops (ICMEW)","volume":"63 5 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2019-07-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128016253","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"Motion-Let Clustering for Skeleton-Based Action Recognition","authors":"Jianyu Yang, Chen Zhu, Junsong Yuan","doi":"10.1109/ICMEW.2019.00033","DOIUrl":"https://doi.org/10.1109/ICMEW.2019.00033","url":null,"abstract":"The representation of 3D skeleton is critical for human action recognition. In this paper, we propose a new approach for skeleton-based action recognition by motion-let clustering. To discover discriminative action features, action sequences are represented by the 3D movements of joints among neighbor frames, called motion-lets. We divide the joints into four groups according to the root-to-end structure, and project the motion-lets to 2D projections. In each 2D projection, the 2D motion-lets of each group are clustered to make primitive representation of actions as histograms. The temporal pyramid is employed to capture the temporal ordering of motion-lets followed by the Random Forest algorithm for classification. The proposed method is validated on four benchmark 3D human action datasets and the experimental results show that the proposed approach outperforms state-of-the-art deep-learning-based methods.","PeriodicalId":302235,"journal":{"name":"2019 IEEE International Conference on Multimedia & Expo Workshops (ICMEW)","volume":"10 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2019-07-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130873437","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"Semantic Segmentation of Satellite Images Using a U-Shaped Fully Connected Network with Dense Residual Blocks","authors":"Eric Narciso Molina, Zenghui Zhang","doi":"10.1109/ICMEW.2019.00037","DOIUrl":"https://doi.org/10.1109/ICMEW.2019.00037","url":null,"abstract":"Semantic segmentation is the task of clustering pixels into an object class. In the field of remote sensing semantic segmentation has wide applications ranging from scene cover classification to change detection for scene understanding. With the success of deep learning algorithms for classification tasks, there has been much work to apply convolutional neural networks in remote sensing with much success. However, feature extraction of high resolution remote sensing imagery poses a challenge when applying such networks. In particular, there is a need to extract high level features while maintaining an objects resolution in the networks feature space. This work proposes an efficient deep fully convolution architecture that obtains high level features without loss of spatial resolution by replacing the standard convolutional layers in U-Net with dense residual blocks. By stacking identity blocks, we allow the input to flow through the network at every proceeding layer. Our network is termed DRU-Net, and is shown to outperform standard U-Net.","PeriodicalId":302235,"journal":{"name":"2019 IEEE International Conference on Multimedia & Expo Workshops (ICMEW)","volume":"21 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2019-07-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114976673","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"Spatial-Temporal Codec Accuracy Calibration for Multi-scale Giga-Pixel Macroscope","authors":"Lei Wang, J. Suo, Jingtao Fan","doi":"10.1109/ICMEW.2019.00077","DOIUrl":"https://doi.org/10.1109/ICMEW.2019.00077","url":null,"abstract":"The video-rate giga-pixel imaging macroscope with centimeter scale field of view and micron resolution, is an indispensable tool for the study of system biology. Its camera array based intrinsic nature brings a significant increase in throughput, but also challenges the applicability and accuracy of existing camera calibration methods. Here we proposed a spatial-temporal coded accuracy calibration method for multi-scale giga-pixel macroscope, which can achieve 0.199 pixels re-projection error. Meanwhile, a vignetting correction is also performed, for the seamless stitching under the condition of multi-camera brightness response anisotropy.","PeriodicalId":302235,"journal":{"name":"2019 IEEE International Conference on Multimedia & Expo Workshops (ICMEW)","volume":"3 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2019-07-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130107856","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"Toward Robust Online Adaptive Visual Tracking via Pyramidal Features Extraction","authors":"Shuai Bai, Yuan Dong, Ting-Bing Xu, Hongliang Bai","doi":"10.1109/ICMEW.2019.00102","DOIUrl":"https://doi.org/10.1109/ICMEW.2019.00102","url":null,"abstract":"Visual object tracking in a complex situation is a challenging problem. In recent years, some high-quality trackers of convolutional neural networks (CNN) have received extensive attention. However, they still suffer from the model drift and limited appearance representations. In this paper, we design a more robust online adaptation tracking framework, called OPT, which consists of pyramidal feature extraction and domain-specific semantic branches. Specifically, we extract hierarchical features from a pre-trained backbone CNN to construct the subsequent pyramidal feature representations. Meanwhile, each domain-specific semantic branch for each sequence from tracking dataset is integrated into the proposed framework to implement the end-to-end training process. To obtain the robust tracking result during the test stage, we introduce an extra domain-specific semantic branch only based on the more distinct shallow features to adaptively update parameters online for the specific tracking target. Additionally, for the same input, the multiple different sample strategies are adopted to form a hard negative balanced sample distribution to achieve sufficient training on the limited data. Finally, extensive experiments on the OTB benchmarks demonstrate the effectiveness and robustness of our proposed tracking framework.","PeriodicalId":302235,"journal":{"name":"2019 IEEE International Conference on Multimedia & Expo Workshops (ICMEW)","volume":"11 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2019-07-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"121617856","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"Deep Key Clips-Video Feature Fusion Framework for Action Recognition","authors":"Chaoyu Li, Yue Ming, Yuan Shen, Hui Yu","doi":"10.1109/ICMEW.2019.00034","DOIUrl":"https://doi.org/10.1109/ICMEW.2019.00034","url":null,"abstract":"Action recognition is crucial for many computer vision applications. Recently, deep learning has made breakthrough in recognition performance of action. However, there are a large number of redundant video frames which contain similar information making it difficult to capture discriminative spatio-temporal features for long-term actions. In this paper, we propose a novel framework for action recognition: Deep Key Clips-Video feature fusion framework. First, we propose a key clip selection algorithm based on background subtraction, which utilizes image average gradient and select key clips for training. Then, we further superimpose the key frames to generate historical contour images, effectively aggregating long-term information of the actions. Key video clips and historical contour images are inputted to the 3D convolutional network and the 2D convolutional network respectively, which extract the clip level and long term video level feature. Finally, we fuse these two sub-networks to improve the accuracy of recognition. We conduct experiments on two current mainstream action recognition datasets UCF-101 and HMDB-51. Compared with the state-of-the-art methods, the experimental results demonstrate the effectiveness of our proposed network for action recognition.","PeriodicalId":302235,"journal":{"name":"2019 IEEE International Conference on Multimedia & Expo Workshops (ICMEW)","volume":"151 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2019-07-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"134313342","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"Scale-Adaptive CNN Based Crowd Counting and Dynamic Supervision","authors":"Zhengxin Li, Jing Li, Ling Xie, Jianli Liu","doi":"10.1109/ICMEW.2019.00076","DOIUrl":"https://doi.org/10.1109/ICMEW.2019.00076","url":null,"abstract":"A major problem in crowd counting is the varying size of human heads in an image due to the changeable camera viewpoint, perspective and imaging resolution. To resolve this problem, we propose scale-adaptive CNN-based framework (SA-CNN) which implicitly outputs features at multiple scales and adaptively estimates density maps from the fusion of these features. To overcome the fact that head size is not annotated in all the public crowd datasets, we design a dynamic supervision strategy to reach the goal. Experiment results show that our method achieve state-of-the-art performance while demonstrating significant robustness to hyper-parameters even the choice of backbone.","PeriodicalId":302235,"journal":{"name":"2019 IEEE International Conference on Multimedia & Expo Workshops (ICMEW)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2019-07-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116327745","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"A Multimedia Based Movie Style Model","authors":"Priyanka Choudhary, Neeraj Goel, Mukesh Saini","doi":"10.1109/ICMEW.2019.0-108","DOIUrl":"https://doi.org/10.1109/ICMEW.2019.0-108","url":null,"abstract":"Movie style is one abstract representation of a movie and is used in recommendation systems. In this paper, we argue that movie style model is more than just emotional attribute represented by genre, but also includes visual, auditory and compositional attributes. We propose a movie model that takes all these attributes into account and generate a comprehensive movie model by automatically analyzing the contents of the movie. The proposed model does not require manual tagging and is independent of the length of the movie. Using a case study of a set of Hollywood directors we show that the attributes proposed by us are important to model a movie style. We show that predictor using these features recognize the correct director with an accuracy of 76.4%.","PeriodicalId":302235,"journal":{"name":"2019 IEEE International Conference on Multimedia & Expo Workshops (ICMEW)","volume":"76 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2019-07-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128530557","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}