🤖 Image AI Renamer

Paste an image → AI recognizes it → file is automatically named and saved. No manual renaming needed.

An Obsidian plugin that automatically calls AI APIs when you paste an image, extracts text (OCR) or classifies image content, and uses the result as a semantic filename.

---

Features

• Auto OCR — Extracts text from screenshots and uses it as the filename
• Auto Image Classification — Recognizes photo content (objects, scenes, categories) when no text is detected
• Smart Fallback — In Auto mode, tries OCR first; falls back to image classification if no text is found
• Keyword Filtering — Strips stop words and meaningless tokens to generate clean, readable filenames
• Multiple Providers — Supports Baidu OCR, Baidu Image Classification, Aliyun, and any custom API endpoint
• Settings UI — Configure your API keys and preferences directly inside Obsidian

---

Demo

Paste a screenshot of a document:

Before: paste → Pasted image 20240101.png
After:  paste → 大学物理_嵌入式系统_人工智能基础_1712345678.png

Paste a photo:

Before: paste → Pasted image 20240101.png
After:  paste → 猫咪_橙色_室内_1712345678.png

---

Installation

From Community Plugins (Recommended)

1. Open Obsidian → Settings → Community plugins
2. Disable Safe Mode if prompted
3. Click Browse and search for Image AI Renamer
4. Click Install → Enable

Manual Installation

1. Download main.js, manifest.json, and styles.css from the latest release
2. Copy them to your vault at .obsidian/plugins/obsidian-image-ai/
3. Reload Obsidian and enable the plugin under Community plugins

---

Setup

Baidu AI (Recommended for Chinese users)

Baidu offers 500 free calls/day for both OCR and image classification.

1. Go to ai.baidu.com and log in
2. Navigate to Console → 文字识别 (OCR) → Create an application
3. Copy the API Key and Secret Key
4. In Obsidian: Settings → Image AI Renamer → paste your keys
5. Make sure the following services are enabled in your Baidu console:
   • 通用文字识别（高精度版）— General OCR (High Accuracy)
   • 图像内容理解 — Image Content Understanding

Aliyun

1. Go to market.aliyun.com and search for 图像识别
2. Purchase a plan (free trials available)
3. Copy your AppCode from the purchased services page
4. Paste it into Settings → Image AI Renamer → AppCode

Custom API

Your endpoint must accept a POST request with:

{
  "image": "<base64 data URI>",
  "mime_type": "image/png"
}

And return one of the following formats:

{ "description": "keyword1_keyword2" }
{ "keywords": ["keyword1", "keyword2"] }
{ "words_result": [{ "words": "recognized text" }] }

---

Recognition Modes (Baidu only)

Mode	Behavior
🔄 Auto (Recommended)	OCR first; falls back to image classification if no text found
📝 OCR only	Always uses text recognition — best for screenshots and documents
🖼️ Classify only	Always uses image classification — best for photos

---

Settings

Setting	Description
AI Provider	Choose between Baidu, Aliyun, or Custom
API Key / Secret Key	Credentials for Baidu OCR
AppCode	Credentials for Aliyun
Custom API URL	Endpoint URL for your own AI service
Custom API Key	Optional Bearer token for your custom endpoint
Recognition Mode	Auto / OCR only / Classify only (Baidu only)
Attachment Folder	Where to save pasted images (default: attachments)

---

How It Works

1. You paste an image in any Obsidian note (edit mode)
2. The plugin intercepts the paste event
3. Sends the image (base64) to your configured AI API
4. Parses the response and extracts meaningful keywords
5. Generates a filename like 关键词A_关键词B_timestamp.png
6. Saves the image to your attachment folder
7. Inserts a Markdown image link into your note

---

Troubleshooting

Paste has no effect

• Confirm the plugin is enabled under Settings → Community plugins
• Make sure you are in Edit mode, not Preview mode

❌ API 连接失败: status 401

• Your API Key or Secret Key is incorrect — copy-paste them again, watch for extra spaces

❌ 服务未开通

• Log in to the Baidu console and manually activate the required services

❌ 图片中未检测到文字

• The image has no readable text; switch to Auto or Classify only mode

Free quota exceeded

• Baidu allows 500 free calls per day; the count resets at midnight (Beijing time)

---

Privacy

• Images are sent to the AI provider you configure (Baidu / Aliyun / your own server)
• No data is collected by this plugin itself
• Your API keys are stored locally in your vault's .obsidian/plugins/obsidian-image-ai/data.json

---

License

MIT © Your Name