Videokapitel extrahieren

Kapitel aus einem Video als strukturiertes JSON extrahieren

Sie können ein Modell direkt abfragen und die zurückgegebenen Ergebnisse testen, wenn Sie verschiedene Parameterwerte mit der Cloud Console oder durch direkten Aufruf der Vertex AI API verwenden.

Freiform-Prompt

Sie können den Videoinhalt in Kapitel unterteilen, indem Sie den Videoinhalt in Kapitel gruppiert und für jedes Kapitel eine Zusammenfassung zusammenfassen. Erfassen Sie nur wichtige Ereignisse und Highlights. Wenn Sie sich bei bestimmten Informationen nicht sicher sind, denken Sie sich bitte nichts aus. Geben Sie das Ergebnis im JSON-Format mit Schlüsseln so zurück: "timecode", "ctxSummary".

Antwort


[
  {
    "timecode": "00:00-00:16",
    "chapterSummary": "The video opens with breathtaking aerial views of Rio de Janeiro, showcasing iconic landmarks such as Sugarloaf Mountain and the Christ the Redeemer statue. It then transitions to scenes of the vibrant beach culture and the bustling city life."
  },
  {
    "timecode": "00:16-00:36",
    "chapterSummary": "The video shifts its focus to the favelas, highlighting the stark contrast between the affluent areas of Rio and these underserved communities. It emphasizes the negative stereotypes associated with favelas, often portrayed through news reports as hubs of crime, poverty, and violence."
  },
  {
    "timecode": "00:36-00:57",
    "chapterSummary": "The video introduces Google's initiative to collaborate with favela residents in mapping their communities. It underscores the importance of providing favelas with a sense of identity and recognition by giving them addresses and making them visible on maps. The chapter emphasizes that favelas are not just places but communities with rich cultures and resilient people."
  }
]
Modell: gemini-1.5-pro-preview-0409
Temperatur: 1
Max. Ausgabetokens: 8.192
TopK: 40
TopP: 0,95