173 research outputs found
A multi-agent 3d simulation environment for clothing industry
No full textThe clothing artefact business is facing relevant restructuring to become able to produce items with enhanced value as for quality reliability, fashion inventiveness and mass customization. The paper presents a multi-agent simulation environment developed to assess and virtually check the feasibility and performances of flexible automation solutions that can help the clothing industry to overcome the shift towards knowledge driven organizations. It addresses new options based on distributed intelligence and robotized cooperative resources including human assisted working
In the Footsteps of Galileo. History of Science in Italian TV Films and Series in the Nineteen-Sixties and Seventies
No full textItalian scientific biopics experienced a period of extraordinary media hype in the 1970s, when some intellectuals personally committed to bringing the lives of the scientists of the past to television in order to discuss the relationship between knowledge and power in the present. Nevertheless, might we properly speak of "Italian-style" historical-scientific fictional drama? To answer this question, we will focus on Roberto Rossellini, Liliana Cavani and, above all, Lucio Lombardo Radice, a promoter, scientific consultant, author and presenter of, and sometimes even actor in, some of the most controversial of these scientific biopics. This article aims, first of all, to reconstruct this history, explaining the reasons for the success of the genre, starting in the 1960s, and the crisis it underwent in the 1980s; secondly, to ascertain the influences these ideological works exerted on choices, approaches and styles of the next generation of science historians and communicators
Dante Gabriel Rossetti "en rapport" with Dante
No full textThis article deals with a "curious incident" in the history of translation studies. In 1909, a translation scholar, William Guthrie, theorises the possibility of reconstructing a Source Text through translation only, without knowledge of the source text itself. He identifies in Dante Gabriel Rossetti's translation an example of this "magic" capacity. This article reconstructs the specific historical context of the "curious incident", pointing at an alleged spiritual contact between the translator and the ST's author, as if they were linked by a mesmeric "rapport"
Profiling Hate Speech Spreaders on Twitter
No full textTask
Hate speech (HS) is commonly defined as any communication that disparages a person or a group on the basis of some characteristic such as race, colour, ethnicity, gender, sexual orientation, nationality, religion, or other characteristics. Given the huge amount of user-generated contents on Twitter, the problem of detecting, and therefore possibly contrasting the HS diffusion, is becoming fundamental, for instance for fighting against misogyny and xenophobia. To this end, in this task, we aim at identifying possible hate speech spreaders on Twitter as a first step towards preventing hate speech from being propagated among online users.
After having addressed several aspects of author profiling in social media from 2013 to 2020 (fake news spreaders, bot detection, age and gender, also together with personality, gender and language variety, and gender from a multimodality perspective), this year we aim at investigating if it is possible to discriminate authors that have shared some hate speech in the past from those that, to the best of our knowledge, have never done it.
As in previous years, we propose the task from a multilingual perspective:
English
Spanish
NOTE: Although we recommend participating in both languages (English and Spanish), it is possible to address the problem just for one language.
Award
We are happy to announce that the best performing team at the 9th International Competition on Author Profiling will be awarded 300,- Euro sponsored by Symanto
Data
Input
The uncompressed dataset consists of a folder per language (en, es). Each folder contains:
An XML file per author (Twitter user) with 100 tweets. The name of the XML file corresponding to the unique author id.
A truth.txt file with the list of authors and the ground truth.
The format of the XML files is:
Tweet 1 textual contents
Tweet 2 textual contents
...
The format of the truth.txt file is as follows. The first column corresponds to the author id. The second column contains the truth label.
b2d5748083d6fdffec6c2d68d4d4442d:::0
2bed15d46872169dc7deaf8d2b43a56:::0
8234ac5cca1aed3f9029277b2cb851b:::1
5ccd228e21485568016b4ee82deb0d28:::0
60d068f9cafb656431e62a6542de2dc0:::1
...
Output
Your software must take as input the absolute path to an unpacked dataset, and has to output for each document of the dataset a corresponding XML file that looks like this:
<author id="author-id"
lang="en|es"
type="0|1"
/>
The naming of the output files is up to you. However, we recommend using the author-id as filename and "XML" as an extension.
IMPORTANT! Languages should not be mixed. A folder should be created for each language and place inside only the files with the prediction for this language.
Evaluation
The performance of your system will be ranked by accuracy. For each language, we will calculate individual accuracies in discriminating between the two classes. Finally, we will average the accuracy values per language to obtain the final ranking.
Related Work
[1] Valerio Basile, Cristina Bosco, Elisabetta Fersini, Dora Nozza, Viviana Patti, Francisco Rangel, Paolo Rosso, Manuela Sanguinetti (2019). SemEval-2019 task 5: Multilingual detection of hate speech against immigrants and women in Twitter. Proc. SemEval 2019
[2] Fabio Poletto, Valerio Basile, Manuela Sanguinetti, Cristina Bosco, Viviana Patti (2020). Resources and benchmark corpora for hate speech detection: a systematic review. Language Resources & Evaluation. https://doi.org/10.1007/s10579-020-09502-8
[3] Paula Fortuna, Sérgio Nunes (2018). A survey on automatic detection of hate speech in text. ACM Computing Surveys (CSUR) 51.4
[4] Maria Anzovino, Elisabetta Fersini, Paolo Rosso (2018). Automatic Identification and Classification of Misogynistic Language on Twitter. In: Proc. 23rd Int. Conf. on Applications of Natural Language to Information Systems, NLDB-2018, Springer-Verlag, LNCS(10859), pp. 57-64
[5] Elisabetta Fersini, Paolo Rosso, Maria Anzovino (2018). Overview of the task on automatic misogyny identification at IberEval 2018. Proc. IberEval 2018
[6] Elisabetta Fersini, Dora Nozza, Paolo Rosso (2018). Overview of the Evalita 2018 task on automatic misogyny identification (AMI). Proc. EVALITA 2018
[7] Cristina Bosco, Felice Dell'Orletta, Fabio Poletto, Manuela Sanguinetti, Maurizio Tesconi (2018). Overview of the EVALITA 2018 hate speech detection task. Proc. EVALITA 2018
[8] Samuel Caetano da Silva, Thiago Castro Ferreira, Ricelli Moreira Silva Ramos, Ivandre Paraboni (2020). Data-driven and psycholinguistics motivated approaches to hate speech detection. Computación y Sistemas, 24(3): 1179–1188
[9] Stiven Zimmerman, Udo Kruschwitz, Cris Fox (2018). Improving hate speech detection with deep learning ensembles. In Proc. of the Eleventh Int. Conf. on Language Resources and Evaluation (LREC 2018)
[10] Francisco Rangel, Anastasia Giachanou, Bilal Ghanem, Paolo Rosso. Overview of the 8th Author Profiling Task at PAN 2020: Profiling Fake News Spreaders on Twitter. In: L. Cappellato, C. Eickhoff, N. Ferro, and A. Névéol (eds.) CLEF 2020 Labs and Workshops, Notebook Papers. CEUR Workshop Proceedings.CEUR-WS.org, vol. 2696
[11] Francisco Rangel and Paolo Rosso. Overview of the 7th Author Profiling Task at PAN 2019: Bots and Gender Profiling in Twitter. In: L. Cappellato, N. Ferro, D. E. Losada and H. Müller (eds.) CLEF 2019 Labs and Workshops, Notebook Papers. CEUR Workshop Proceedings.CEUR-WS.org, vol. 2380
[12] Francisco Rangel, Paolo Rosso, Martin Potthast, Benno Stein. Overview of the 6th author profiling task at pan 2018: multimodal gender identification in Twitter. In: CLEF 2018 Labs and Workshops, Notebook Papers. CEUR Workshop Proceedings. CEUR-WS.org, vol. 2125.
[13] Francisco Rangel, Paolo Rosso, Martin Potthast, Benno Stein. Overview of the 5th Author Profiling Task at PAN 2017: Gender and Language Variety Identification in Twitter. In: Cappellato L., Ferro N., Goeuriot L, Mandl T. (Eds.) CLEF 2017 Labs and Workshops, Notebook Papers. CEUR Workshop Proceedings. CEUR-WS.org, vol. 1866.
[14] Francisco Rangel, Paolo Rosso, Ben Verhoeven, Walter Daelemans, Martin Pottast, Benno Stein. Overview of the 4th Author Profiling Task at PAN 2016: Cross-Genre Evaluations. In: Balog K., Capellato L., Ferro N., Macdonald C. (Eds.) CLEF 2016 Labs and Workshops, Notebook Papers. CEUR Workshop Proceedings. CEUR-WS.org, vol. 1609, pp. 750-784
[15] Francisco Rangel, Fabio Celli, Paolo Rosso, Martin Pottast, Benno Stein, Walter Daelemans. Overview of the 3rd Author Profiling Task at PAN 2015.In: Linda Cappelato and Nicola Ferro and Gareth Jones and Eric San Juan (Eds.): CLEF 2015 Labs and Workshops, Notebook Papers, 8-11 September, Toulouse, France. CEUR Workshop Proceedings. ISSN 1613-0073, http://ceur-ws.org/Vol-1391/,2015.
[16] Francisco Rangel, Paolo Rosso, Irina Chugur, Martin Potthast, Martin Trenkmann, Benno Stein, Ben Verhoeven, Walter Daelemans. Overview of the 2nd Author Profiling Task at PAN 2014. In: Cappellato L., Ferro N., Halvey M., Kraaij W. (Eds.) CLEF 2014 Labs and Workshops, Notebook Papers. CEUR-WS.org, vol. 1180, pp. 898-827.
[17] Francisco Rangel, Paolo Rosso, Moshe Koppel, Efstatios Stamatatos, Giacomo Inches. Overview of the Author Profiling Task at PAN 2013. In: Forner P., Navigli R., Tufis D. (Eds.)Notebook Papers of CLEF 2013 LABs and Workshops. CEUR-WS.org, vol. 1179
[18] Francisco Rangel and Paolo Rosso On the Implications of the General Data Protection Regulation on the Organisation of Evaluation Tasks. In: Language and Law / Linguagem e Direito, Vol. 5(2), pp. 80-102
[19] Francisco Rangel, Marc Franco-Salvador, Paolo Rosso A Low Dimensionality Representation for Language Variety Identification. In: Postproc. 17th Int. Conf. on Comput. Linguistics and Intelligent Text Processing, CICLing-2016, Springer-Verlag, Revised Selected Papers, Part II, LNCS(9624), pp. 156-169 (arXiv:1705.10754
Dante Gabriel Rossetti "en rapport" with Dante
No full textThis article deals with a "curious incident" in the history of translation studies. In 1909, a translation scholar, William Guthrie, theorises the possibility of reconstructing a Source Text through translation only, without knowledge of the source text itself. He identifies in Dante Gabriel Rossetti's translation an example of this "magic" capacity. This article reconstructs the specific historical context of the "curious incident", pointing at an alleged spiritual contact between the translator and the ST's author, as if they were linked by a mesmeric "rapport"
The ancient library of the bolognese convent of S. Paolo in Monte (l’Osservanza): catalogue of manuscripts
No full textIl lavoro è dedicato allo studio della collezione manoscritta di epoca medievale appartenuta al convento Osservante bolognese di S. Paolo in Monte, fondato agli inizi del Quattrocento. Il primo capitolo ripercorre la storia dell’ente, dalla fondazione all’epoca moderna, presentando numerose fonti storico-documentarie riemerse sia dall’archivio del convento stesso (attualmente conservato presso l’Archivio di Stato di Bologna), sia dalla tradizione storiografica francescana a partire dal XVI secolo. Il secondo capitolo pone l’attenzione sulla libraria del convento, la cui esistenza risulta già attestata intorno alla metà del XV secolo, cercando di ricostruire le vicende genetiche ed evolutive della collezione attraverso l’indagine di documenti d’archivio, fonti catalografiche e sopravvivenze manoscritte. Vengono presentate, in particolare, due liste librarie relative a questa biblioteca: la prima, già nota alla bibliografia più recente, fu redatta agli inizi del XVI secolo da un umanista italiano di nome Fabio Vigili (92 item) ed edita nel 1943 da padre M. J. Laurent nell’ambito di una più ampia serie di cataloghi di biblioteche emiliano-romagnole (M. H. LAURENT, Fabio Vigili et les bibliothèques de Bologne au debut du XVIe siècle d’après le Ms. Barb. Lat. 3185, Studi e testi, 105) Città del Vaticano, 1943); la seconda, esemplata nel 1600 nell’ambito dell’inchiesta della S. Congregazione dell’Indice, e mai attenzionata è stata, invece, recentemente riscoperta da chi scrive all’interno del ms. Città del Vaticano, BAV, Vat. lat. 11271 (ff. 81r-98v). Di entrambe le liste librarie si offre la trascrizione integrale, mettendone in luce le concordanze. Attualmente sono stati individuati 65 manoscritti appartenuti al convento di S. Paolo in Monte, dislocati in diverse sedi di conservazione dei quali si è approntato un catalogo descrittivo che costituisce la parte più consistente del lavoro. Chiudono il lavoro un’ampia serie di tavole, e la bibliografia di riferimento.This study investigates the medieval manuscript collection of the Franciscan Observant convent of San Paolo in Monte (also known as l’Osservanza), founded in Bologna in the early 15th century. The first chapter traces the history of the convent from its foundation to the modern era, drawing upon a wide range of historical and documentary sources—many of which have recently come to light in the convent’s own archive (now housed at the Archivio di Stato di Bologna) and within the broader Franciscan historiographical tradition from the 16th century onwards. The second chapter focuses on the convent’s library, already attested by the mid-15th century. Through archival documents, catalog records, and surviving manuscripts, the study reconstructs the origins and development of the collection. Two key book lists are examined: the first, compiled in the early 16th century by the Italian humanist Fabio Vigili (92 items), was published by M. J. Laurent in 1943; the second, previously unknown, was rediscovered by the author in Vatican Library, Vat. lat. 11271 (ff. 81r–98v), and dates to 1600. The study includes full transcriptions of both inventories, with commentary and concordance analysis. It further traces the fate of the library through the Napoleonic (1810) and post-Unification (1866) suppressions, which led to the collection’s dispersal. To date, 65 manuscripts from San Paolo in Monte have been identified in modern repositories—including the Biblioteca Universitaria and Archiginnasio in Bologna, the Houghton Library at Harvard, and the Sacro Convento in Assisi. The core of the work is a descriptive catalogue of these manuscripts, enriched with historical and codicological insights. The volume concludes with a comprehensive set of tables and a full scholarly bibliography
VISIONE Feature Repository for VBS: Multi-Modal Features and Detected Objects from MVK Dataset
No full text<p>This repository contains a diverse set of features extracted from the marine video (underwater) dataset (MVK) . These features were utilized in the VISIONE system [Amato et al. 2023, Amato et al. 2022] during the latest editions of the Video Browser Showdown (VBS) competition (<a href="https://www.videobrowsershowdown.org/">https://www.videobrowsershowdown.org/</a>). </p>
<p>We used a snapshot of the MVK dataset from 2023, that can be downloaded using the instructions provided at <a href="https://download-dbis.dmi.unibas.ch/mvk/">https://download-dbis.dmi.unibas.ch/mvk/</a>. It comprises 1,372 video files. We divided each video into 1 second segments. </p>
<p>This repository is released under a Creative Commons Attribution license. If you use it in any form for your work, please cite the following paper:</p>
<blockquote>
<pre>@inproceedings{amato2023visione,
title={VISIONE at Video Browser Showdown 2023},
author={Amato, Giuseppe and Bolettieri, Paolo and Carrara, Fabio and Falchi, Fabrizio and Gennaro, Claudio and Messina, Nicola and Vadicamo, Lucia and Vairo, Claudio},
booktitle={International Conference on Multimedia Modeling},
pages={615--621},
year={2023},
organization={Springer}
}</pre>
</blockquote>
<p> </p>
<p>This repository comprises the following files:</p>
<ul>
<li><strong><em>msb.tar.gz </em></strong> contains tab-separated files (.tsv) for each video. Each tsv file reports, for each video segment, the timestamp and frame number marking the start/end of the video segment, along with the timestamp of the extracted middle frame and the associated identifier ("id_visione"). </li>
<li><em><strong>extract-keyframes-from-msb.tar.gz</strong></em> contains a Python script designed to extract the middle frame of each video segment from the MSB files. To run the script successfully, please ensure that you have the original MVK videos available.</li>
<li><strong><em>features-aladin.tar.gz<sup>†</sup></em> </strong>contains <a href="https://github.com/mesnico/ALADIN">ALADIN</a> [Messina N. et al. 2022] features extracted for all the segment's middle frames. </li>
<li><em><strong>features-clip-laion.tar.gz<sup>†</sup></strong></em> contains <a href="https://huggingface.co/laion/CLIP-ViT-H-14-laion2B-s32B-b79K">CLIP ViT-H/14 - LAION-2B </a>[Schuhmann et al. 2022] features extracted for all the segment's middle frames.</li>
<li><em><strong>features-clip-openai.tar.gz<sup>†</sup> </strong></em>contains <a href="https://huggingface.co/openai/clip-vit-large-patch14">CLIP ViT-L/14</a> [Radford et al. 2021] features extracted for all the segment's middle frames. </li>
<li><em><strong>features-clip2video.tar.gz<sup>†</sup> </strong></em>contains <a href="https://github.com/CryhanFang/CLIP2Video">CLIP2Video</a> [Fang H. et al. 2021] extracted for all the 1s video segments. <strong> </strong></li>
<li><em><strong>objects-frcnn-oiv4.tar.gz<sup>*</sup> </strong></em>contains the objects detected using <a href="http://tfhub.dev/google/faster_rcnn/openimages_v4/inception_resnet_v2/1">Faster R-CNN+Inception ResNet</a> (trained on the Open Images V4 [Kuznetsova et al. 2020]). </li>
<li><em><strong>objects-mrcnn-lvis.tar.gz<sup>*</sup></strong></em> contains the objects detected using Mask R-CNN [He et al. 2017] (trained on LVIS).</li>
<li><em><strong>objects-vfnet64-coco.tar.gz<sup>*</sup></strong></em> contains the objects detected using VfNet [Zhang et al. 2021] (trained on COCO dataset).</li>
</ul>
<p>*Please be sure to use the <strong>v2 version </strong>of this repository, since v1 feature files may contain inconsistencies that have now been corrected</p>
<p><em><strong>*Note on the object annotations:</strong></em> Within an object archive, there is a jsonl file for each video, where each row contains a record of a video segment (the <em>"_id"</em> corresponds to the <em>"id_visione"</em> used in the msb.tar.gz) . Additionally, there are three arrays representing the objects detected, the corresponding scores, and the bounding boxes. The format of these arrays is as follows:</p>
<ul>
<li><em>"object_class_names"</em>: vector with the class name of each detected object.</li>
<li><em>"object_scores"</em>: scores corresponding to each detected object.</li>
<li><em>"object_boxes_yxyx"</em>: bounding boxes of the detected objects in the format <em>(ymin, xmin, ymax, xmax).</em></li>
</ul>
<p> </p>
<p><em><strong><sup>†</sup>Note on the cross-modal features: </strong></em>The extracted multi-modal features (ALADIN, CLIPs, CLIP2Video) enable internal searches within the MVK dataset using the query-by-image approach (features can be compared with the dot product). However, to perform searches based on free text, the text needs to be transformed into the joint embedding space according to the specific network being used (see links above). Please be aware that t<strong>he service for transforming text into features is not provided within this repository and should be developed independently using the original feature repositories linked above.</strong></p>
<p>We have plans to release the code in the future, allowing the reproduction of the VISIONE system, including the instantiation of all the services to transform text into cross-modal features. However, this work is still in progress, and the code is not currently available.</p>
<p> </p>
<p><strong>References:</strong></p>
<p>[Amato et al. 2023] Amato, G.et al., 2023, January. VISIONE at Video Browser Showdown 2023. In International Conference on Multimedia Modeling (pp. 615-621). Cham: Springer International Publishing.</p>
<p>[Amato et al. 2022] Amato, G. et al. (2022). VISIONE at Video Browser Showdown 2022. In: , et al. MultiMedia Modeling. MMM 2022. Lecture Notes in Computer Science, vol 13142. Springer, Cham. </p>
<p>[Fang H. et al. 2021] Fang H. et al., 2021. Clip2video: Mastering video-text retrieval via image clip. arXiv preprint arXiv:2106.11097.</p>
<p>[He et al. 2017] He, K., Gkioxari, G., Dollár, P. and Girshick, R., 2017. Mask r-cnn. In Proceedings of the IEEE international conference on computer vision (pp. 2961-2969).</p>
<p>[Kuznetsova et al. 2020] Kuznetsova, A., Rom, H., Alldrin, N., Uijlings, J., Krasin, I., Pont-Tuset, J., Kamali, S., Popov, S., Malloci, M., Kolesnikov, A. and Duerig, T., 2020. The open images dataset v4. International Journal of Computer Vision, 128(7), pp.1956-1981.</p>
<p>[Lin et al. 2014] Lin, T.Y., Maire, M., Belongie, S., Hays, J., Perona, P., Ramanan, D., Dollár, P. and Zitnick, C.L., 2014, September. Microsoft coco: Common objects in context. In European conference on computer vision (pp. 740-755). Springer, Cham.</p>
<p>[Messina et al. 2022] Messina N. et al., 2022, September. Aladin: distilling fine-grained alignment scores for efficient image-text matching and retrieval. In Proceedings of the 19th International Conference on Content-based Multimedia Indexing (pp. 64-70).</p>
<p>[Radford et al. 2021] Radford A. et al., 2021, July. Learning transferable visual models from natural language supervision. In International conference on machine learning (pp. 8748-8763). PMLR.</p>
<p>[Schuhmann et al. 2022] Schuhmann C. et al., 2022. Laion-5b: An open large-scale dataset for training next generation image-text models. Advances in Neural Information Processing Systems, 35, pp.25278-25294.</p>
<p>[Zhang et al. 2021] Zhang, H., Wang, Y., Dayoub, F. and Sunderhauf, N., 2021. Varifocalnet: An iou-aware dense object detector. In Proceedings of the IEEE/CV</p>
VISIONE Feature Repository for VBS: Multi-Modal Features and Detected Objects from V3C1+V3C2 Dataset
No full text<p>This repository contains a diverse set of features extracted from the V3C1+V3C2 dataset, sourced from the Vimeo Creative Commons Collection. These features were utilized in the VISIONE system [Amato et al. 2023, Amato et al. 2022] during the latest editions of the Video Browser Showdown (VBS) competition (<a href="https://www.videobrowsershowdown.org/">https://www.videobrowsershowdown.org/</a>).</p>
<p>The original V3C1+V3C2 dataset, provided by NIST, can be downloaded using the instructions provided at <a href="https://videobrowsershowdown.org/about-vbs/existing-data-and-tools/">https://videobrowsershowdown.org/about-vbs/existing-data-and-tools/</a>.</p>
<p>It comprises 7,235 video files, amounting for 2,300h of video content and encompassing 2,508,113 predefined video segments.</p>
<p>We subdivided the predefined video segments longer than 10 seconds into multiple segments, with each segment spanning no longer than 16 seconds. As a result, we obtained a total of 2,648,219 segments. For each segment, we extracted one frame, specifically the middle one, and computed several features, which are described in detail below.</p>
<p>This repository is released under a Creative Commons Attribution license. If you use it in any form for your work, please cite the following paper:</p>
<blockquote>
<pre>@inproceedings{amato2023visione,
title={VISIONE at Video Browser Showdown 2023},
author={Amato, Giuseppe and Bolettieri, Paolo and Carrara, Fabio and Falchi, Fabrizio and Gennaro, Claudio and Messina, Nicola and Vadicamo, Lucia and Vairo, Claudio},
booktitle={International Conference on Multimedia Modeling},
pages={615--621},
year={2023},
organization={Springer}
}</pre>
</blockquote>
<p> </p>
<p>This repository comprises the following files:</p>
<ul>
<li><strong><em>msb.tar.gz </em></strong> contains tab-separated files (.tsv) for each video. Each tsv file reports, for each video segment, the timestamp and frame number marking the start/end of the video segment, along with the timestamp of the extracted middle frame and the associated identifier ("id_visione"). </li>
<li><em><strong>extract-keyframes-from-msb.tar.gz</strong></em> contains a Python script designed to extract the middle frame of each video segment from the MSB files. To run the script successfully, please ensure that you have the original V3C videos available.</li>
<li><strong><em>features-aladin.tar.gz<sup>†</sup></em> </strong>contains <a href="https://github.com/mesnico/ALADIN">ALADIN</a> [Messina N. et al. 2022] features extracted for all the segment's middle frames. </li>
<li><em><strong>features-clip-laion.tar.gz<sup>†</sup></strong></em> contains <a href="https://huggingface.co/laion/CLIP-ViT-H-14-laion2B-s32B-b79K">CLIP ViT-H/14 - LAION-2B </a>[Schuhmann et al. 2022] features extracted for all the segment's middle frames.</li>
<li><em><strong>features-clip-openai.tar.gz<sup>†</sup> </strong></em>contains <a href="https://huggingface.co/openai/clip-vit-large-patch14">CLIP ViT-L/14</a> [Radford et al. 2021] features extracted for all the segment's middle frames. </li>
<li><em><strong>features-clip2video.tar.gz<sup>†</sup> </strong></em>contains <a href="https://github.com/CryhanFang/CLIP2Video">CLIP2Video</a> [Fang H. et al. 2021] extracted for all the video segments. <strong> </strong>In particular 1) we concatenate consecutive short segments so to create segments at least 3 seconds long; 2) we downsample the obtained segments to 2.5 fps; 3) we feed the network with the first min(36, n) frames, where n is the number of frames of the segment. Notice that the minimum processed length consists of 7 frames, given that the segment is no shorter than 3s. </li>
<li><em><strong>objects-frcnn-oiv4.tar.gz<sup>*</sup> </strong></em>contains the objects detected using <a href="http://tfhub.dev/google/faster_rcnn/openimages_v4/inception_resnet_v2/1">Faster R-CNN+Inception ResNet</a> (trained on the Open Images V4 [Kuznetsova et al. 2020]). </li>
<li><em><strong>objects-mrcnn-lvis.tar.gz<sup>*</sup></strong></em> contains the objects detected using Mask R-CNN [He et al. 2017] (trained on LVIS).</li>
<li><em><strong>objects-vfnet64-coco.tar.gz<sup>*</sup></strong></em> contains the objects detected using VfNet [Zhang et al. 2021] (trained on COCO dataset).</li>
</ul>
<p>*Please be sure to use the <strong>v2 version </strong>of this repository, since v1 feature files may contain inconsistencies that have now been corrected</p>
<p><em><strong>*Note on the object annotations:</strong></em> Within an object archive, there is a jsonl file for each video, where each row contains a record of a video segment (the <em>"_id"</em> corresponds to the <em>"id_visione"</em> used in the msb.tar.gz) . Additionally, there are three arrays representing the objects detected, the corresponding scores, and the bounding boxes. The format of these arrays is as follows:</p>
<ul>
<li><em>"object_class_names"</em>: vector with the class name of each detected object.</li>
<li><em>"object_scores"</em>: scores corresponding to each detected object.</li>
<li><em>"object_boxes_yxyx"</em>: bounding boxes of the detected objects in the format <em>(ymin, xmin, ymax, xmax).</em></li>
</ul>
<p> </p>
<p><em><strong><sup>†</sup>Note on the cross-modal features: </strong></em>The extracted multi-modal features (ALADIN, CLIPs, CLIP2Video) enable internal searches within the V3C1+V3C2 dataset using the query-by-image approach (features can be compared with the dot product). However, to perform searches based on free text, the text needs to be transformed into the joint embedding space according to the specific network being used. Please be aware that t<strong>he service for transforming text into features is not provided within this repository and should be developed independently using the original feature repositories linked above.</strong></p>
<p>We have plans to release the code in the future, allowing the reproduction of the VISIONE system, including the instantiation of all the services to transform text into cross-modal features. However, this work is still in progress, and the code is not currently available.</p>
<p> </p>
<p><strong>References:</strong></p>
<p>[Amato et al. 2023] Amato, G.et al., 2023, January. VISIONE at Video Browser Showdown 2023. In International Conference on Multimedia Modeling (pp. 615-621). Cham: Springer International Publishing.</p>
<p>[Amato et al. 2022] Amato, G. et al. (2022). VISIONE at Video Browser Showdown 2022. In: , et al. MultiMedia Modeling. MMM 2022. Lecture Notes in Computer Science, vol 13142. Springer, Cham. </p>
<p>[Fang H. et al. 2021] Fang H. et al., 2021. Clip2video: Mastering video-text retrieval via image clip. arXiv preprint arXiv:2106.11097.</p>
<p>[He et al. 2017] He, K., Gkioxari, G., Dollár, P. and Girshick, R., 2017. Mask r-cnn. In Proceedings of the IEEE international conference on computer vision (pp. 2961-2969).</p>
<p>[Kuznetsova et al. 2020] Kuznetsova, A., Rom, H., Alldrin, N., Uijlings, J., Krasin, I., Pont-Tuset, J., Kamali, S., Popov, S., Malloci, M., Kolesnikov, A. and Duerig, T., 2020. The open images dataset v4. International Journal of Computer Vision, 128(7), pp.1956-1981.</p>
<p>[Lin et al. 2014] Lin, T.Y., Maire, M., Belongie, S., Hays, J., Perona, P., Ramanan, D., Dollár, P. and Zitnick, C.L., 2014, September. Microsoft coco: Common objects in context. In European conference on computer vision (pp. 740-755). Springer, Cham.</p>
<p>[Messina et al. 2022] Messina N. et al., 2022, September. Aladin: distilling fine-grained alignment scores for efficient image-text matching and retrieval. In Proceedings of the 19th International Conference on Content-based Multimedia Indexing (pp. 64-70).</p>
<p>[Radford et al. 2021] Radford A. et al., 2021, July. Learning transferable visual models from natural language supervision. In International conference on machine learning (pp. 8748-8763). PMLR.</p>
<p>[Schuhmann et al. 2022] Schuhmann C. et al., 2022. Laion-5b: An open large-scale dataset for training next generation image-text models. Advances in Neural Information Processing Systems, 35, pp.25278-25294.</p>
<p>[Zhang et al. 2021] Zhang, H., Wang, Y., Dayoub, F. and Sunderhauf, N., 2021. Varifocalnet: An iou-aware dense object detector. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (pp. 8514-8523).</p>
The language of Poesie a Casarsa. The "invented" friulian of Pier Paolo Pasolini
No full textopenIl presente elaborato analizza il “timp furlan” di Pier Paolo Pasolini, con particolare attenzione rivolta alla pubblicazione di Poesie a Casarsa (1942) e alla lingua utilizzata dall’autore per comporre le poesie della raccolta. La ricerca prende avvio con un rapido excursus sugli sviluppi della poesia dialettale tra XIX e XX secolo, approfondendo l’influenza esercitata da Pascoli sui poeti dialettali di primo Novecento e il tentativo di conferire dignità linguistica e letteraria ai dialetti. Il cuore della trattazione concentra l’interesse sull’esperienza trascorsa in Friuli di Pasolini e sulla lingua friulana intesa dal poeta come lingua per la poesia, l’unica capace di toccare le corde più profonde dell’animo del poeta e di portare alla memoria la purezza degli anni trascorsi in Friuli. Lo studio prosegue approfondendo lo sviluppo sia strutturale che linguistico di Poesie a Casarsa, dal progetto de I confini in italiano, alla dialetizzazione dei componimenti, sino alla pubblicazione della raccolta in dialetto casarsese. La veste linguistica della prima raccolta friulana non dimostra una totale fedeltà al dialetto casarsese poiché accoglie elementi derivanti da friulani differenti. Viene proposta quindi una analisi linguistica del dialetto utilizzato da Pasolini, mostrando le variazioni che allontanano la lingua della raccolta dal reale casarsese e proponendo successivamente la corretta terminologia.This paper analyses Pier Paolo Pasolini's "timp furlan", with particular attention paid to the pubblication of Poesie a Casarsa (1942) and the language used by the author to compose the poems in the collection. This research begins with a quick excursus on the development of dialect poetry between the 19th and 20th centuries, examining in depth the influence exercised by Pascoli on early 20th century dialect poets and the attempt to confer linguitic and literary dignity on dialects. The heart of the treatment focuses on Pasolini's experience in Friuli and on the Friulian language, understood by the poet as a language of poetry, the only one capable of touching che deepest chords of the poet's soul and bringing to memory the purity of years spent in Friuli. The study continues by delving into both the structural and linguistic development of Poesie a Casarsa, from the project of I confini in italian, to the dialetisation of the poems, up to the collection in Casarsa dialect. The linguistic guise of the first friulian collection does not demostrare total fidelity to the dialect of Casarsa, as it incorporates elements derivig from different friulian dialects. Therefore, a linguistic analysis of the dialect used by Pasolini in proposed, showing the variations that distance the language of the collection from the real Casarsese and then proposing the correct terminology
- …
