[et_pb_section fb_built=”1″ background_color=”#070028″ fullwidth=”on” prev_background_color=”#ffffff” next_background_color=”#ffffff” _builder_version=”3.0.105″ top_divider_style=”slant2″ bottom_divider_style=”arrow2″][et_pb_fullwidth_header title=”Publications” text_orientation=”center” _builder_version=”3.0.105″ title_font=”||||||||” title_text_color=”#ffffff”][\/et_pb_fullwidth_header][\/et_pb_section][et_pb_section fb_built=”1″ _builder_version=”3.0.105″ custom_padding=”54px|0px|0px|0px”][et_pb_row make_fullwidth=”on” _builder_version=”3.0.105″][et_pb_column type=”1_3″ _builder_version=”3.0.105″ parallax=”off” parallax_method=”on”][et_pb_blurb image=”http:\/\/4dbeyond.eu\/wp-content\/uploads\/2021\/01\/elidek_logo.png” _builder_version=”3.0.105″ animation=”off”][\/et_pb_blurb][\/et_pb_column][et_pb_column type=”1_3″ _builder_version=”3.0.105″ parallax=”off” parallax_method=”on”][et_pb_blurb image=”http:\/\/4dbeyond.eu\/wp-content\/uploads\/2021\/01\/gsrt_logo_focus.png” _builder_version=”3.0.105″ animation=”off”][\/et_pb_blurb][\/et_pb_column][et_pb_column type=”1_3″ _builder_version=”3.0.105″ parallax=”off” parallax_method=”on”][et_pb_blurb image=”http:\/\/4dbeyond.eu\/wp-content\/uploads\/2021\/01\/ntua.jpg” image_max_width=”36%” _builder_version=”3.0.105″ animation=”off”][\/et_pb_blurb][\/et_pb_column][\/et_pb_row][\/et_pb_section][et_pb_section fb_built=”1″ fullwidth=”on” _builder_version=”3.0.105″ background_color=”#070028″ bottom_divider_style=”mountains2″ bottom_divider_color=”#ffffff” bottom_divider_height=”53px”][et_pb_fullwidth_header subhead=”The research work was supported by the Hellenic Foundation for Research and Innovation (H.F.R.I.) under the \u201cFirst Call for H.F.R.I. Research Projects to support Faculty members and Researchers and the procurement of high-cost research equipment grant\u201d (Project Number: FRI-FM17-2972).” _builder_version=”3.0.105″ background_layout=”dark” box_shadow_style=”preset1″ title_font=”||||||||” title_text_shadow_style=”preset3″ title_text_shadow_color=”rgba(255,255,255,0.4)”][\/et_pb_fullwidth_header][\/et_pb_section][et_pb_section fb_built=”1″ background_color=”#070028″ next_background_color=”#ffffff” _builder_version=”3.0.105″ top_divider_style=”arrow3″ bottom_divider_style=”mountains2″ custom_padding=”55px|0px|54px|0px” top_divider_color=”#ffffff” top_divider_height=”15px”][et_pb_row _builder_version=”3.0.105″][et_pb_column type=”4_4″ _builder_version=”3.0.105″ parallax=”off” parallax_method=”on”][et_pb_post_slider show_pagination=”off” show_meta=”off” _builder_version=”3.0.105″ background_color=”#070028″ box_shadow_style=”preset1″ box_shadow_blur=”23px” box_shadow_spread=”19px” box_shadow_color=”rgba(255,255,255,0.3)”][\/et_pb_post_slider][\/et_pb_column][\/et_pb_row][\/et_pb_section][et_pb_section fb_built=”1″ _builder_version=”3.0.105″][et_pb_row _builder_version=”3.0.105″][et_pb_column type=”1_3″ _builder_version=”3.0.105″ parallax=”off” parallax_method=”on”][et_pb_blurb title=”Unsupervised 3D Motion Summarization Using Stacked Auto-Encoders” _builder_version=”3.0.105″ text_orientation=”justified”]<\/p>\n
Abstract<\/strong>:\u00a0In this paper, a deep stacked auto-encoder (SAE) scheme followed by a hierarchical Sparse Modeling for Representative Selection (SMRS) algorithm is proposed to summarize dance video sequences, recorded using the VICON Motion capturing system. SAE\u2019s main task is to reduce the redundant information embedding in the raw data and, thus, to improve summarization performance. This becomes apparent when two dancers are performing simultaneously and severe errors are encountered in the humans\u2019 point joints, due to dancers\u2019 occlusions in the 3D space. Four summarization algorithms are applied to extract the key frames; density based, Kennard Stone, conventional SMRS and its hierarchical scheme called H-SMRS. Experimental results have been carried out on real-life dance sequences of Greek traditional dances while the results have been compared against ground truth data selected by dance experts. The results indicate that H-SMRS being applied after the SAE information reduction module extracts key frames which are deviated in time less than 0.3 s to the ones selected by the experts and with a standard deviation of 0.18 s. Thus, the proposed scheme can effectively represent the content of the dance sequence<\/span><\/p>\n <\/span><\/p>\n [\/et_pb_blurb][\/et_pb_column][et_pb_column type=”1_3″ _builder_version=”3.0.105″ parallax=”off” parallax_method=”on”][et_pb_blurb title=”Adaptable Autoregressive Moving AverageFilter Triggering Convolutional Neural Networks for Choreographic Modelling” _builder_version=”3.0.105″ text_orientation=”justified”]<\/p>\n Abstract<\/strong>:\u00a0Choreographic modeling, that is identification of key choreographic primitives, is a significant element for Intangible Cultural<\/span>Heritage (ICH) performing art modeling. Recently, deep learning architectures, such as LSTM and CNN, have been utilized for<\/span>choreographic identification and modeling. However, such approaches present sensitivity to capturing errors and fail to model the<\/span>dynamic characteristics of a dance, since they assume a stationarity between the input-output data. To address these limitations,<\/span>in this paper, we introduce an AutoRegressive Moving Average (ARMA) filter into a conventional CNN model; this means that<\/span>the classification output feeds back to the input layer, improving overall classification accuracy. In addition, an adaptive imple-<\/span>mentation algorithm is introduced, exploiting a first-order Taylor series expansion, to update network response in order to fit dance<\/span>dynamic characteristics. This way, the network parameters (e.g., weights) are dynamically modified improving overall classification<\/span>accuracy. Experimental results on real-life dance sequences indicate the out-performance of the proposed approach with respect to<\/span>conventional deep learning mechanisms.<\/span><\/p>\n https:\/\/www.researchgate.net\/publication\/343400674_ADAPTABLE_AUTOREGRESSIVE_MOVING_AVERAGE_FILTER_TRIGGERING_CONVOLUTIONAL_NEURAL_NETWORKS_FOR_CHOREOGRAPHIC_MODELING<\/a><\/span><\/p>\n <\/span>https:\/\/www.isprs-ann-photogramm-remote-sens-spatial-inf-sci.net\/V-2-2020\/467\/2020\/isprs-annals-V-2-2020-467-2020.pdf<\/span><\/a><\/p>\n [\/et_pb_blurb][\/et_pb_column][et_pb_column type=”1_3″ _builder_version=”3.0.105″ parallax=”off” parallax_method=”on”][et_pb_blurb title=”Automatic 3D Modeling and Reconstruction of Cultural Heritage Sites from Twitter Images” _builder_version=”3.0.105″ text_orientation=”justified”]<\/p>\n Abstract<\/strong>:\u00a0This paper presents an approach for leveraging the abundance of images posted on social media like Twitter for large scale 3D reconstruction of cultural heritage landmarks. Twitter allows users to post short messages, including photos, describing a plethora of activities or events, e.g., tweets are used by travelers on vacation, capturing images from various cultural heritage assets. As such, a great number of images are available online, able to drive a successful 3D reconstruction process. However, reconstruction of any asset, based on images mined from Twitter, presents several challenges. There are three main steps that have to be considered: (i) tweets\u2019 content identification, (ii) image retrieval and filtering, and (iii) 3D reconstruction. The proposed approach first extracts key events from unstructured tweet messages and then identifies cultural activities and landmarks. The second stage is the application of a content-based filtering method so that only a small but representative portion of cultural images are selected to support fast 3D reconstruction. The proposed methods are experimentally evaluated using real-world data and comparisons verify the effectiveness of the proposed scheme<\/p>\n https:\/\/www.mdpi.com\/2071-1050\/12\/10\/4223\/htm<\/a><\/a><\/span><\/p>\n