apple/ml-hypersim

• hightopics#1

  Add relevant topics to the repository

  Why:

  COPY-PASTE FIX

  Copy

  synthetic-data, computer-vision, scene-understanding, indoor-scenes, per-pixel-labels, photorealistic, dataset, machine-learning, deep-learning

  OpenDoneDismiss
• highreadme#2

  Reposition README H1 and opening paragraph for clarity

  Why:

  CURRENT

  # The Hypersim Dataset
  
  For many fundamental scene understanding tasks, it is difficult or impossible to obtain per-pixel ground truth labels from real images. We address this challenge by introducing Hypersim, a photorealistic synthetic dataset for holistic indoor scene understanding.

  COPY-PASTE FIX

  Copy

  # The Hypersim Dataset: A Photorealistic Synthetic Dataset for Holistic Indoor Scene Understanding
  
  Hypersim is a large-scale, photorealistic synthetic dataset specifically designed for computer vision and robotics researchers working on holistic indoor scene understanding. It addresses the challenge of obtaining per-pixel ground truth labels from real images by providing 77,400 images of 461 indoor scenes with detailed per-pixel labels and corresponding ground truth geometry.

  OpenDoneDismiss
• mediumreadme#3

  Create a dedicated 'Key Features' section in the README

  Why:

  CURRENT

  Our dataset: (1) relies exclusively on publicly available 3D assets; (2) includes complete scene geometry, material information, and lighting information for every scene; (3) includes dense per-pixel semantic instance segmentations and complete camera information for every image; and (4) factors every image into diffuse reflectance, diffuse illumination, and a non-diffuse residual term that captures view-dependent lighting effects.

  COPY-PASTE FIX

  Copy

  ## Key Features
  
  Our Hypersim dataset offers:
  
  *   **Publicly Available Assets:** Relies exclusively on publicly available 3D assets.
  *   **Complete Scene Information:** Includes complete scene geometry, material information, and lighting information for every scene.
  *   **Dense Per-Pixel Labels:** Provides dense per-pixel semantic instance segmentations and complete camera information for every image.
  *   **Factored Images:** Factors every image into diffuse reflectance, diffuse illumination, and a non-diffuse residual term that captures view-dependent lighting effects.

  OpenDoneDismiss

Category GEO backends resolved for this scan: google/gemini-2.5-flash, deepseek/deepseek-v4-flash

• CATEGORY QUERY

  Where can I find a synthetic dataset with per-pixel labels for indoor scene understanding?

  you: not recommended
  AI recommended (in order):

  1. Matterport3D
  2. ScanNet
  3. SUN RGB-D
  4. Habitat-Matterport3D
  5. Replica
  6. Gibson
  7. NYU Depth V2

  AI recommended 7 alternatives but never named apple/ml-hypersim. This is the gap to close.

  Show full AI answer
• CATEGORY QUERY

  What are good photorealistic synthetic datasets for training computer vision models on indoor environments?

  you: not recommended
  AI recommended (in order):

  1. Replica Dataset
  2. Matterport3D
  3. Habitat-Matterport3D
  4. ScanNet
  5. Hypersim

  AI recommended 5 alternatives but never named apple/ml-hypersim. This is the gap to close.

  Show full AI answer

• Metadata completeness

  warn

  Suggestion:

• README presence

  pass

• Compared to common alternatives in this category, what is the core differentiator of apple/ml-hypersim?

  pass
  ✓AI named apple/ml-hypersim explicitly

  AI answers can be confidently wrong. Read for accuracy: does it match your actual tech stack, audience, and differentiator?

• If a team adopts apple/ml-hypersim in production, what risks or prerequisites should they evaluate first?

  pass
  ✓AI named apple/ml-hypersim explicitly

  AI answers can be confidently wrong. Read for accuracy: does it match your actual tech stack, audience, and differentiator?

• In one sentence, what problem does the repo apple/ml-hypersim solve, and who is the primary audience?

  pass
  ⚠AI did not name apple/ml-hypersim — likely talking about a different project

  AI answers can be confidently wrong. Read for accuracy: does it match your actual tech stack, audience, and differentiator?