{
  "docId": "019dd923-5e88-73ef-bd5d-0856e1444fb9",
  "docSlug": "8757f1b44ef7f176",
  "documentTitle": "2024 Air Street Capital The State of AI Report 2024",
  "authorId": "AirStreetCapital",
  "authorName": "Air Street Capital",
  "documentKindSlug": "consulting-deck",
  "documentKindLabel": "Consulting deck",
  "sourceTypeSlug": "vc_research",
  "sourceTypeLabel": "VC research",
  "presentationDate": null,
  "orientation": "landscape",
  "aspectRatio": 1.777,
  "pageNumber": 73,
  "pageCount": 213,
  "prevPage": 72,
  "nextPage": 74,
  "slideType": "industry_trends",
  "function": "summarize",
  "density": "overcrowded",
  "nDataPoints": 4,
  "notes": "Part of the State of AI 2024 report.",
  "elementsJson": [
    "headline_text",
    "bullet_list",
    "photo"
  ],
  "metadataConfidence": 1,
  "imagePath": null,
  "slideHref": "/slides/019dd923-5e88-73ef-bd5d-0856e1444fb9/73",
  "deckHref": "/decks/019dd923-5e88-73ef-bd5d-0856e1444fb9",
  "deckJsonHref": "/decks/019dd923-5e88-73ef-bd5d-0856e1444fb9.json",
  "deckAnchorHref": "/decks/019dd923-5e88-73ef-bd5d-0856e1444fb9#slide-73",
  "components": [
    {
      "bbox": null,
      "kind": "callout",
      "text": "They find it is more accurate and 6x faster the SAM 1 at image segmentation, while able to surpass the accuracy of prior leading video segmentation models with 3x fewer interactions.",
      "attrs": null,
      "subkind": null,
      "toolName": "Visual emphasis",
      "toolSlug": "visual-emphasis",
      "confidence": null,
      "componentId": "019dd952-47c3-7407-8d15-f1b02bfe018e",
      "frameworkName": null,
      "frameworkSlug": null
    },
    {
      "bbox": {
        "h": 0.6,
        "w": 0.3,
        "x": 0.65,
        "y": 0.3
      },
      "kind": "image",
      "text": "Visual examples of SAM 2 segmentation",
      "attrs": null,
      "subkind": "infographic",
      "toolName": null,
      "toolSlug": null,
      "confidence": null,
      "componentId": "b1891003-dd25-4668-bd34-dc31709f51a7",
      "frameworkName": null,
      "frameworkSlug": null
    },
    {
      "bbox": {
        "h": 0.1,
        "w": 0.6,
        "x": 0.02,
        "y": 0.6
      },
      "kind": "list",
      "text": "They find it is more accurate and 6x faster the SAM 1 at image segmentation, while able to surpass the accuracy of prior leading video segmentation models with 3x fewer interactions.",
      "attrs": null,
      "subkind": "bullet",
      "toolName": null,
      "toolSlug": null,
      "confidence": null,
      "componentId": "2b78f829-c042-4607-9d92-237e8503be67",
      "frameworkName": null,
      "frameworkSlug": null
    },
    {
      "bbox": {
        "h": 0.15,
        "w": 0.6,
        "x": 0.02,
        "y": 0.45
      },
      "kind": "list",
      "text": "To build a unified model that works for both video as well as image, Meta have made some adaptations. For example, they included a memory mechanism to track objects across frames and an occlusion head to handle objects that disappear or reappear.",
      "attrs": null,
      "subkind": "bullet",
      "toolName": null,
      "toolSlug": null,
      "confidence": null,
      "componentId": "3790b35d-4f21-4a13-88bb-5efc99af2a79",
      "frameworkName": null,
      "frameworkSlug": null
    },
    {
      "bbox": {
        "h": 0.15,
        "w": 0.6,
        "x": 0.02,
        "y": 0.3
      },
      "kind": "list",
      "text": "Meta has extended SAM to include video segmentation, training it on their own dataset (SA-V) of 51,000 real-world videos and 600,000 spatio-temporal masks. This dataset has been made available, along with the model, under an Apache 2.0 license.",
      "attrs": null,
      "subkind": "bullet",
      "toolName": null,
      "toolSlug": null,
      "confidence": null,
      "componentId": "51a9d004-17ca-4445-9457-47732fdddea8",
      "frameworkName": null,
      "frameworkSlug": null
    },
    {
      "bbox": {
        "h": 0.08,
        "w": 0.6,
        "x": 0.02,
        "y": 0.7
      },
      "kind": "list",
      "text": "The model is, however, less efficient at segmenting multiple objects simultaneously in video and can struggle with longer clips.",
      "attrs": null,
      "subkind": "bullet",
      "toolName": null,
      "toolSlug": null,
      "confidence": null,
      "componentId": "e0566a05-ee18-49e6-b29a-08fc4299f7b8",
      "frameworkName": null,
      "frameworkSlug": null
    },
    {
      "bbox": null,
      "kind": "metric",
      "text": "processing speed: 6x",
      "attrs": null,
      "subkind": "primary",
      "toolName": "Quantification",
      "toolSlug": "quantification",
      "confidence": null,
      "componentId": "019dd952-47c3-7407-8d15-f45efc2a3a87",
      "frameworkName": null,
      "frameworkSlug": null
    },
    {
      "bbox": {
        "h": 0.1,
        "w": 0.6,
        "x": 0.02,
        "y": 0.2
      },
      "kind": "paragraph",
      "text": "Last year's Meta's Segment Anything impressed with its ability to identify and segment objects in images given any prompt. In July, they released Segment Anything 2 (SAM 2), which stunned observers.",
      "attrs": null,
      "subkind": "paragraph",
      "toolName": null,
      "toolSlug": null,
      "confidence": null,
      "componentId": "14122076-c28e-4941-93eb-51140080e356",
      "frameworkName": null,
      "frameworkSlug": null
    },
    {
      "bbox": {
        "h": 0.04,
        "w": 0.5,
        "x": 0.02,
        "y": 0.14
      },
      "kind": "title",
      "text": "Segment Anything gets boosters and expands to video",
      "attrs": null,
      "subkind": "headline",
      "toolName": null,
      "toolSlug": null,
      "confidence": null,
      "componentId": "cd50accb-dc5c-47e9-875c-487d2122f525",
      "frameworkName": null,
      "frameworkSlug": null
    }
  ],
  "metrics": [],
  "tools": [
    {
      "name": "Comparison frame",
      "slug": "comparison-frame",
      "agent": null,
      "layer": "slide",
      "matchId": "8e6329a7-7dd1-4097-9071-fee398a7c88f",
      "evidence": "the components compare SAM 1 and SAM 2, highlighting improvements in accuracy and speed",
      "confidence": 0.7
    }
  ],
  "frameworks": [],
  "arcBeats": [
    {
      "to": 153,
      "from": 9,
      "beatId": "019dd95a-0682-776c-8e35-5f2398b8d1d0",
      "arcName": "The Triple Take",
      "arcSlug": "triple-take",
      "beatName": "The Facts (What)",
      "beatSlug": "triple-take-the-facts-what",
      "evidence": "Exec summary + Research + Industry sections inventory the year",
      "position": 1,
      "confidence": 78,
      "parentBeatName": "Setup",
      "parentBeatSlug": "setup"
    },
    {
      "to": 85,
      "from": 12,
      "beatId": "019dd95a-0682-776c-8e35-6d9d8ee06d0a",
      "arcName": "Voyage and Return",
      "arcSlug": "voyage-return",
      "beatName": "The Unknown",
      "beatSlug": "voyage-return-the-unknown",
      "evidence": "Research section explores frontier model uncharted territory",
      "position": 2,
      "confidence": 55,
      "parentBeatName": "Complication",
      "parentBeatSlug": "complication"
    }
  ],
  "loops": [
    {
      "to": 81,
      "from": 72,
      "name": "Pattern Hunter",
      "slug": "02-pattern-hunter",
      "bestFor": "Time-pressed audiences, building consensus, when data is strong",
      "matchId": "019dd95a-07fe-70ce-8d3e-431790a1bfd7",
      "evidence": "Wayve LINGO-2, SAM2, DeepMind robotics, LeRobot, diffusion policies, humanoids, Vision Pro",
      "position": 9,
      "objective": "Show robotics renaissance via converging evidence",
      "structure": "Evidence A -> Evidence B -> Evidence C -> Pattern/Conclusion",
      "confidence": 78,
      "description": "Group multiple pieces of evidence that together point to a pattern or conclusion"
    }
  ],
  "imagePathAlt": null,
  "thumbSrc": null,
  "thumbSrcAlt": null,
  "locked": true
}