La interfaz está localizada; las descripciones y documentos de las Skills se traducen automáticamente con IA a su idioma. El código, los comandos, rutas y términos técnicos se mantienen en el original, y prevalece el documento del autor.
Explorar Skills
Datos e investigación · SKILL.md

image-evaluation

image-evaluation

Provides reference for evaluating AI-generated images and text-to-image (T2I) models. Covers mainstream benchmarks (PartiPrompt, GenAI-Bench, VQAScore, ImageEval, DrawBench, T2I-CompBench++, HRSBench, RAHF, BodyMetric, Human Distortion, Reason-Edit, etc.), metrics (FID, CLIPScore, VQAScore, human alignment, realism), and dataset scales. Use when the user as…

Independent GitHub snapshot · Automated classification · Original documentation governs

Documentación de la Skill

GitHub

name: image-evaluation description: | Provides reference for evaluating AI-generated images and text-to-image (T2I) models. Covers mainstream benchmarks (PartiPrompt, GenAI-Bench, VQAScore, ImageEval, DrawBench, T2I-CompBench++, HRSBench, RAHF, BodyMetric, Human Distortion, Reason-Edit, etc.), metrics (FID, CLIPScore, VQAScore, human alignment, realism), and dataset scales. Use when the user asks about AI image evaluation, T2I benchmarks, image quality metrics, 图像评估, 文生图评测, benchmark, VQAScore, FID, 图文对齐, or how to evaluate generated images. license: MIT metadata: author: 葱姜蒜 version: "1.0.0" homepage: https://github.com/ppdbxdawj/ai-skills tags: - evaluation - t2i - benchmark - image-quality - text-image-alignment

Image Evaluation Skill | 图像评估技能

Helps choose and apply AI image / T2I evaluation benchmarks and metrics. Full benchmark details (metrics, datasets, links) are in reference.md.

协助选择和应用 AI 图像 / 文生图评测 的基准与指标。各基准的指标、数据集与资料链接见 reference.md。

When to use | 何时使用

  • 需要评估文生图/生成图像质量、图文对齐、人体真实感、组合能力等
  • 选 benchmark(如 GenAI-Bench、PartiPrompt、ImageEval、T2I-CompBench++)
  • 选指标(VQAScore、FID、CLIPScore、人评、热力图/可解释反馈)
  • 查数据集量级与评测方式

Quick reference | 速查

需求推荐
通用图文对齐GenAI-Bench + VQAScore
组合/推理T2I-CompBench++、HRSBench、DALL-EVAL
中文 + 细粒度ImageEval
人体真实感/失真BodyMetric、Human Distortion Benchmark
可解释与改进RAHF(热力图 + 分数)
指令编辑Reason-Edit

Single-image evaluation (unified report)

Given one image and one prompt, run multi-metric evaluation and get one HTML report with:

维度指标说明
图文相关性CLIPScore0–1
图文相关性VQAScoreGenAI-Bench 风格,BLIP-2,可选 --no-vqa
图文相关性ImageReward人类偏好奖励模型,与人工评判更一致,可选 --no-imagereward
生成质量锐度 Laplacian无参考清晰度 0–1
美学/吸引力CLIP 美学代理0–1
人体真实感/失真/RAHF—需专用模型,见 reference.md
cd image-evaluation
python3 -m venv .venv && .venv/bin/pip install -r requirements-eval.txt
.venv/bin/python evaluate_image.py --image /path/to/image.png "生图 prompt" --output-dir reports
# 可选:--no-vqa 跳过 VQAScore;--no-imagereward 跳过 ImageReward

报告输出到 reports/report_YYYY-MM-DD_HH-MM-SS.html(每次唯一)。布局:顶部粘性摘要 + 左图右表,减少纵向滚动。

References

  • Full benchmarks & metrics → reference.md(总览表 + 各基准说明 + 使用建议)
Skills guardadas