ssundaram21/dreamsim

• highreadme#1

  Reposition the README's opening paragraph for direct clarity

  Why:

  CURRENT

  Current metrics for perceptual image similarity operate at the level of pixels and patches. These metrics compare images in terms of their low-level colors and textures, but fail to capture mid-level differences in layout, pose, semantic content, etc. Models that use image-level embeddings such as DINO and CLIP capture high-level and semantic judgements, but may not be aligned with human perception of more finegrained attributes. DreamSim is a new metric for perceptual image similarity that bridges the gap between "low-level" metrics (e.g. LPIPS, PSNR, SSIM) and "high-level" measures (e.g. CLIP). Our model was trained by concatenating CLIP, OpenCLIP, and DINO embeddings, and then finetuning on human perceptual judgements. We gathered these judgements on a dataset of ~20k image triplets, generated by diffusion models. Our model achieves better alignment with human similarity judgements than existing metrics, and can be used for downstream applications such as image retrieval.

  COPY-PASTE FIX

  Copy

  DreamSim is a novel metric for perceptual image similarity, specifically designed to align with human judgments, especially for images generated by diffusion models. It bridges the gap between traditional low-level metrics (like LPIPS, PSNR, SSIM) and high-level embedding-based measures (like CLIP and DINO), offering a more human-aligned evaluation for generative AI applications and image retrieval.

  OpenDoneDismiss
• mediumabout#2

  Refine the 'About' description for conciseness

  Why:

  CURRENT

  DreamSim: Learning New Dimensions of Human Visual Similarity using Synthetic Data (NeurIPS 2023 Spotlight) / / / / When Does Perceptual Alignment Benefit Vision Representations? (NeurIPS 2024)

  COPY-PASTE FIX

  Copy

  DreamSim: A human-aligned perceptual similarity metric for images, especially those generated by diffusion models. (NeurIPS 2023 Spotlight & NeurIPS 2024)

  OpenDoneDismiss

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

• CATEGORY QUERY

  Seeking a visual similarity metric that accurately reflects human perception.

  you: not recommended
  AI recommended (in order):

  1. pHash
  2. SSIM
  3. ResNet
  4. VGG
  5. Inception
  6. SIFT
  7. SURF
  8. ORB
  9. Chi-Squared distance
  10. Intersection
  11. Correlation

  AI recommended 11 alternatives but never named ssundaram21/dreamsim. This is the gap to close.

  Show full AI answer
• CATEGORY QUERY

  How to evaluate image generation models with a human-aligned perceptual similarity score?

  you: not recommended
  AI recommended (in order):

  1. CLIP Score (openai/CLIP)
  2. DINO Score
  3. LPIPS (richzhang/PerceptualSimilarity)
  4. FID
  5. KID
  6. PPL (NVlabs/stylegan-xl)
  7. Human Evaluation
  8. Amazon Mechanical Turk
  9. Figure Eight

  AI recommended 9 alternatives but never named ssundaram21/dreamsim. 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 ssundaram21/dreamsim?

  pass
  ✓AI named ssundaram21/dreamsim explicitly

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

• If a team adopts ssundaram21/dreamsim in production, what risks or prerequisites should they evaluate first?

  pass
  ✓AI named ssundaram21/dreamsim 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 ssundaram21/dreamsim solve, and who is the primary audience?

  pass
  ⚠AI did not name ssundaram21/dreamsim — 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?