Conference paper Open Access

Learning Short-Cut Connections for Object Counting

Daniel Oñoro-Rubio; Mathias Niepert; Roberto J. López-Sastre


JSON Export

{
  "files": [
    {
      "links": {
        "self": "https://zenodo.org/api/files/6099e427-89d5-43a1-b14d-9d2226abd7a2/Onoro-Rubio_Learning%20Short-Cut%20Connections%20for%20Object.pdf"
      }, 
      "checksum": "md5:9974e8667992b03179bfee5dfeb4bb28", 
      "bucket": "6099e427-89d5-43a1-b14d-9d2226abd7a2", 
      "key": "Onoro-Rubio_Learning Short-Cut Connections for Object.pdf", 
      "type": "pdf", 
      "size": 1887158
    }
  ], 
  "owners": [
    73409
  ], 
  "doi": "10.5281/zenodo.3813490", 
  "stats": {
    "version_unique_downloads": 12.0, 
    "unique_views": 14.0, 
    "views": 15.0, 
    "version_views": 15.0, 
    "unique_downloads": 12.0, 
    "version_unique_views": 14.0, 
    "volume": 24533054.0, 
    "version_downloads": 13.0, 
    "downloads": 13.0, 
    "version_volume": 24533054.0
  }, 
  "links": {
    "doi": "https://doi.org/10.5281/zenodo.3813490", 
    "conceptdoi": "https://doi.org/10.5281/zenodo.3813489", 
    "bucket": "https://zenodo.org/api/files/6099e427-89d5-43a1-b14d-9d2226abd7a2", 
    "conceptbadge": "https://zenodo.org/badge/doi/10.5281/zenodo.3813489.svg", 
    "html": "https://zenodo.org/record/3813490", 
    "latest_html": "https://zenodo.org/record/3813490", 
    "badge": "https://zenodo.org/badge/doi/10.5281/zenodo.3813490.svg", 
    "latest": "https://zenodo.org/api/records/3813490"
  }, 
  "conceptdoi": "10.5281/zenodo.3813489", 
  "created": "2020-05-07T13:14:12.632738+00:00", 
  "updated": "2020-05-11T20:20:33.043643+00:00", 
  "conceptrecid": "3813489", 
  "revision": 3, 
  "id": 3813490, 
  "metadata": {
    "access_right_category": "success", 
    "doi": "10.5281/zenodo.3813490", 
    "version": "v.2", 
    "language": "akh", 
    "title": "Learning Short-Cut Connections for Object Counting", 
    "license": {
      "id": "CC-BY-4.0"
    }, 
    "related_identifiers": [
      {
        "scheme": "doi", 
        "identifier": "10.5281/zenodo.3813489", 
        "relation": "isVersionOf"
      }
    ], 
    "relations": {
      "version": [
        {
          "count": 1, 
          "index": 0, 
          "parent": {
            "pid_type": "recid", 
            "pid_value": "3813489"
          }, 
          "is_last": true, 
          "last_child": {
            "pid_type": "recid", 
            "pid_value": "3813490"
          }
        }
      ]
    }, 
    "grants": [
      {
        "code": "761508", 
        "links": {
          "self": "https://zenodo.org/api/grants/10.13039/501100000780::761508"
        }, 
        "title": "5GCITY", 
        "acronym": "5GCITY", 
        "program": "H2020", 
        "funder": {
          "doi": "10.13039/501100000780", 
          "acronyms": [], 
          "name": "European Commission", 
          "links": {
            "self": "https://zenodo.org/api/funders/10.13039/501100000780"
          }
        }
      }
    ], 
    "communities": [
      {
        "id": "5gcity"
      }
    ], 
    "publication_date": "2018-11-15", 
    "creators": [
      {
        "affiliation": "NEC Lab Europe", 
        "name": "Daniel O\u00f1oro-Rubio"
      }, 
      {
        "affiliation": "NEC Lab Europe", 
        "name": "Mathias Niepert"
      }, 
      {
        "affiliation": "University de Alcal\u00e1 de Henares", 
        "name": "Roberto J. L\u00f3pez-Sastre"
      }
    ], 
    "meeting": {
      "title": "BMBC 2018"
    }, 
    "access_right": "open", 
    "resource_type": {
      "subtype": "conferencepaper", 
      "type": "publication", 
      "title": "Conference paper"
    }, 
    "description": "<p>Object counting is an important task in computer vision due to its growing demand in applications such as traffic monitoring or surveillance. In this paper, we consider object counting as a learning problem of a joint feature extraction and pixel-wise object density estimation with Convolutional-Deconvolutional networks. We introduce a novel counting model, named Gated U-Net (GU-Net). Specifically, we propose to enrich the U-Net architecture with the concept of learnable short-cut connections. Standard short-cut connections are connections between layers in deep neural networks which skip at least one intermediate layer. Instead of simply setting short-cut connections, we propose to learn these connections from data. Therefore, our short-cuts can work as gating units, which optimize the flow of information between convolutional and deconvolutional layers in the U-Net architecture. We evaluate the introduced GU-Net architecture on three commonly used benchmark data sets for object counting. GU-Nets consistently outperform the base U-Net architecture, and achieve state-of-the-art performance.</p>"
  }
}
15
13
views
downloads
All versions This version
Views 1515
Downloads 1313
Data volume 24.5 MB24.5 MB
Unique views 1414
Unique downloads 1212

Share

Cite as