Conference paper Open Access

Attention Mechanisms, Signal Encodings and Fusion Strategies for Improved Ad-hoc Video Search with Dual Encoding Networks

Galanopoulos, Damianos; Mezaris, Vasileios


JSON Export

{
  "files": [
    {
      "links": {
        "self": "https://zenodo.org/api/files/48749e5f-c44d-49f8-b351-79c82c2af96d/icmr2020_preprint.pdf"
      }, 
      "checksum": "md5:4bd02af5fbd7afd9fd7827bf95a3de67", 
      "bucket": "48749e5f-c44d-49f8-b351-79c82c2af96d", 
      "key": "icmr2020_preprint.pdf", 
      "type": "pdf", 
      "size": 3126217
    }
  ], 
  "owners": [
    22750
  ], 
  "doi": "10.1145/3372278.3390737", 
  "stats": {
    "version_unique_downloads": 22.0, 
    "unique_views": 77.0, 
    "views": 77.0, 
    "version_views": 77.0, 
    "unique_downloads": 22.0, 
    "version_unique_views": 77.0, 
    "volume": 71902991.0, 
    "version_downloads": 23.0, 
    "downloads": 23.0, 
    "version_volume": 71902991.0
  }, 
  "links": {
    "doi": "https://doi.org/10.1145/3372278.3390737", 
    "latest_html": "https://zenodo.org/record/4244549", 
    "bucket": "https://zenodo.org/api/files/48749e5f-c44d-49f8-b351-79c82c2af96d", 
    "badge": "https://zenodo.org/badge/doi/10.1145/3372278.3390737.svg", 
    "html": "https://zenodo.org/record/4244549", 
    "latest": "https://zenodo.org/api/records/4244549"
  }, 
  "created": "2020-11-04T15:05:47.891856+00:00", 
  "updated": "2020-11-05T00:26:57.128588+00:00", 
  "conceptrecid": "4244548", 
  "revision": 3, 
  "id": 4244549, 
  "metadata": {
    "access_right_category": "success", 
    "doi": "10.1145/3372278.3390737", 
    "description": "<p>In this paper, the problem of unlabeled video retrieval using textual queries is addressed. We present an extended dual encoding network which makes use of more than one encodings of the visual and textual content, as well as two different attention mechanisms. The latter serve the purpose of highlighting temporal locations in every modality that can contribute more to effective retrieval. The different encodings of the visual and textual inputs, along with early/late fusion strategies, are examined for further improving performance. Experimental evaluations and comparisons with state-of-the-art methods document the merit of the proposed network.</p>", 
    "license": {
      "id": "CC-BY-4.0"
    }, 
    "title": "Attention Mechanisms, Signal Encodings and Fusion Strategies for Improved Ad-hoc Video Search with Dual Encoding Networks", 
    "relations": {
      "version": [
        {
          "count": 1, 
          "index": 0, 
          "parent": {
            "pid_type": "recid", 
            "pid_value": "4244548"
          }, 
          "is_last": true, 
          "last_child": {
            "pid_type": "recid", 
            "pid_value": "4244549"
          }
        }
      ]
    }, 
    "communities": [
      {
        "id": "retv-h2020"
      }
    ], 
    "grants": [
      {
        "code": "780656", 
        "links": {
          "self": "https://zenodo.org/api/grants/10.13039/501100000780::780656"
        }, 
        "title": "Enhancing and Re-Purposing TV Content for Trans-Vector Engagement", 
        "acronym": "ReTV", 
        "program": "H2020", 
        "funder": {
          "doi": "10.13039/501100000780", 
          "acronyms": [], 
          "name": "European Commission", 
          "links": {
            "self": "https://zenodo.org/api/funders/10.13039/501100000780"
          }
        }
      }
    ], 
    "keywords": [
      "Video search", 
      "Video retrieval", 
      "Ad-hoc video search", 
      "Deep learning", 
      "Dual encoding network", 
      "Attention mechanism"
    ], 
    "publication_date": "2020-06-07", 
    "creators": [
      {
        "affiliation": "CERTH-ITI", 
        "name": "Galanopoulos, Damianos"
      }, 
      {
        "affiliation": "CERTH-ITI", 
        "name": "Mezaris, Vasileios"
      }
    ], 
    "meeting": {
      "acronym": "ICMR 2020", 
      "dates": "2020", 
      "title": "ACM Int. Conf. on Multimedia Retrieval"
    }, 
    "access_right": "open", 
    "resource_type": {
      "subtype": "conferencepaper", 
      "type": "publication", 
      "title": "Conference paper"
    }
  }
}
77
23
views
downloads
Views 77
Downloads 23
Data volume 71.9 MB
Unique views 77
Unique downloads 22

Share

Cite as