使用Chroma和LlamaIndex构建多模态检索系统

在本篇文章中，我们将介绍如何使用Chroma和LlamaIndex构建一个多模态检索系统。Chroma是一个专为开发者生产力和幸福感设计的AI原生开源向量数据库，本教程将展示如何在其中创建多模态索引并进行检索。

安装所需库

首先，安装所需的软件包：

pip install chromadb
pip install llama-index
pip install llama-index-vector-stores-qdrant
pip install llama-index-embeddings-huggingface
pip install llama-index-vector-stores-chroma
pip install sentence-transformers
pip install pydantic==1.10.11
pip install open-clip-torch

创建Chroma索引

首先，我们来创建一个Chroma索引：

# 导入必要的库
from llama_index.core import VectorStoreIndex, SimpleDirectoryReader
from llama_index.vector_stores.chroma import ChromaVectorStore
from llama_index.core import StorageContext
from llama_index.embeddings.huggingface import HuggingFaceEmbedding
from IPython.display import Markdown, display
import chromadb

# 设置OpenAI API Key
import os
import openai

OPENAI_API_KEY = "你的OpenAI API密钥"
openai.api_key = OPENAI_API_KEY
os.environ["OPENAI_API_KEY"] = OPENAI_API_KEY

下载维基百科图片和文本

我们将从维基百科中下载一些文本和图片数据：

import requests
from pathlib import Path
import urllib.request

def get_wikipedia_images(title):
    response = requests.get(
        "https://en.wikipedia.org/w/api.php",
        params={
            "action": "query",
            "format": "json",
            "titles": title,
            "prop": "imageinfo",
            "iiprop": "url|dimensions|mime",
            "generator": "images",
            "gimlimit": "50",
        },
    ).json()
    image_urls = []
    for page in response["query"]["pages"].values():
        if page["imageinfo"][0]["url"].endswith(".jpg") or page["imageinfo"][0]["url"].endswith(".png"):
            image_urls.append(page["imageinfo"][0]["url"])
    return image_urls

image_uuid = 0
MAX_IMAGES_PER_WIKI = 20

wiki_titles = {
    "Tesla Model X",
    "Pablo Picasso",
    "Rivian",
    "The Lord of the Rings",
    "The Matrix",
    "The Simpsons",
}

data_path = Path("mixed_wiki")
if not data_path.exists():
    Path.mkdir(data_path)

for title in wiki_titles:
    response = requests.get(
        "https://en.wikipedia.org/w/api.php",
        params={
            "action": "query",
            "format": "json",
            "titles": title,
            "prop": "extracts",
            "explaintext": True,
        },
    ).json()
    page = next(iter(response["query"]["pages"].values()))
    wiki_text = page["extract"]

    with open(data_path / f"{title}.txt", "w") as fp:
        fp.write(wiki_text)

    images_per_wiki = 0
    try:
        list_img_urls = get_wikipedia_images(title)
        for url in list_img_urls:
            if url.endswith(".jpg") or url.endswith(".png"):
                image_uuid += 1
                urllib.request.urlretrieve(url, data_path / f"{image_uuid}.jpg")
                images_per_wiki += 1
                if images_per_wiki > MAX_IMAGES_PER_WIKI:
                    break
    except:
        print(str(Exception("No images found for Wikipedia page: ")) + title)
        continue

设置嵌入模型

我们需要设置默认的文本和图像嵌入函数：

from chromadb.utils.embedding_functions import OpenCLIPEmbeddingFunction

embedding_function = OpenCLIPEmbeddingFunction()

构建Chroma多模态索引

通过LlamaIndex构建Chroma多模态索引：

from llama_index.core.indices import MultiModalVectorStoreIndex
from llama_index.vector_stores.qdrant import QdrantVectorStore
from chromadb.utils.data_loaders import ImageLoader

image_loader = ImageLoader()

# 创建客户端和新的集合
chroma_client = chromadb.EphemeralClient()
chroma_collection = chroma_client.create_collection(
    "multimodal_collection",
    embedding_function=embedding_function,
    data_loader=image_loader,
)

# 加载文档
documents = SimpleDirectoryReader("./mixed_wiki/").load_data()

# 设置ChromaVectorStore并加载数据
vector_store = ChromaVectorStore(chroma_collection=chroma_collection)
storage_context = StorageContext.from_defaults(vector_store=vector_store)
index = VectorStoreIndex.from_documents(
    documents,
    storage_context=storage_context,
)

从多模态索引中检索结果

最后，我们将从多模态索引中检索结果：

retriever = index.as_retriever(similarity_top_k=50)
retrieval_results = retriever.retrieve("Picasso famous paintings")

from llama_index.core.schema import ImageNode
from llama_index.core.response.notebook_utils import (
    display_source_node,
    display_image_uris,
)

image_results = []
MAX_RES = 5
cnt = 0
for r in retrieval_results:
    if isinstance(r.node, ImageNode):
        image_results.append(r.node.metadata["file_path"])
    else:
        if cnt < MAX_RES:
            display_source_node(r)
        cnt += 1

display_image_uris(image_results, [3, 3], top_k=2)

可能遇到的错误

1. 缺少依赖库：确保所有所需的Python库都已安装。
2. API Key错误：确保你已正确设置了OpenAI API Key。
3. 网络问题：可能会由于网络问题导致无法从维基百科下载数据。
4. 文件路径问题：确保指定的文件路径存在或是可读可写的。

参考资料:

• Chroma Documentation
• LlamaIndex Documentation

如果你觉得这篇文章对你有帮助，请点赞，关注我的博客，谢谢!