{
  "absUrl": "https://arxiv.org/abs/2609.39938",
  "arxivId": "2609.39938",
  "arxivVersion": null,
  "arxivVersionedId": null,
  "authors": [
    {
      "affiliations": [
        "Northeastern University",
        "Futurewei Technologies"
      ],
      "name": "Juyi Lin"
    },
    {
      "affiliations": [
        "Futurewei Technologies"
      ],
      "name": "Zhiqiang Lao"
    },
    {
      "affiliations": [
        "Futurewei Technologies"
      ],
      "name": "Jiali Cui"
    },
    {
      "affiliations": [
        "Northeastern University"
      ],
      "name": "Lin Zhao"
    },
    {
      "affiliations": [
        "Northeastern University"
      ],
      "name": "Pu Zhao"
    },
    {
      "affiliations": [
        "Futurewei Technologies"
      ],
      "name": "Dichang Zhang"
    },
    {
      "affiliations": [
        "Northeastern University"
      ],
      "name": "Arman Akbari"
    },
    {
      "affiliations": [
        "Northeastern University"
      ],
      "name": "Yu Qi"
    },
    {
      "affiliations": [
        "Northeastern University"
      ],
      "name": "Xinru Jiang"
    },
    {
      "affiliations": [
        "Northeastern University"
      ],
      "name": "Yanzhi Wang"
    },
    {
      "affiliations": [
        "Futurewei Technologies"
      ],
      "name": "Heather Yu"
    },
    {
      "affiliations": [
        "Futurewei Technologies"
      ],
      "name": "Liang Peng"
    }
  ],
  "contract": "researcher-sidecars-v1",
  "id": "2609.39938",
  "originalTitle": "LEAP: Learned Block-wise Evidence Retrieval for Long Audio-Video Perception",
  "pdfUrl": "https://arxiv.org/pdf/2609.39938.pdf",
  "schemaVersion": 1,
  "type": "preprint"
}