List out of range error when uploading images.

[nhtkid]

Dear Family,

I am using the image loader as below:

                    image_loader = UnstructuredImageLoader(file_path, mode="single")
                    # Use [0] to get the single Langchain document
                    image_data = image_loader.load()[0]
                    # Append the Langchain document to the documents list
                    documents.append(image_data)

The documents variable will be used for chunking and embedding.

        text_splitter = RecursiveCharacterTextSplitter(chunk_size=1500, chunk_overlap=150)
        document_chunks = text_splitter.split_documents(documents)
        embeddings = OpenAIEmbeddings()
        vectorstore = Chroma.from_documents(document_chunks, embeddings)

The problem now is when I upload an image, I get the below error from Streamlit.
However, in the same code, I also used UnstructuredFileLoader, which I have no problem uploading PDF/TXT files.

Traceback (most recent call last):

  File "/home/adminuser/venv/lib/python3.10/site-packages/streamlit/runtime/scriptrunner/script_runner.py", line 552, in _run_script

    exec(code, module.__dict__)

  File "/mount/src/chatgptwithyourdata/ChatWithYourData_v2.py", line 103, in <module>

    vectorstore = Chroma.from_documents(document_chunks, embeddings)

  File "/home/adminuser/venv/lib/python3.10/site-packages/langchain/vectorstores/chroma.py", line 603, in from_documents

    return cls.from_texts(

  File "/home/adminuser/venv/lib/python3.10/site-packages/langchain/vectorstores/chroma.py", line 567, in from_texts

    chroma_collection.add_texts(texts=texts, metadatas=metadatas, ids=ids)

  File "/home/adminuser/venv/lib/python3.10/site-packages/langchain/vectorstores/chroma.py", line 226, in add_texts

    self._collection.upsert(

  File "/home/adminuser/venv/lib/python3.10/site-packages/chromadb/api/models/Collection.py", line 298, in upsert

    ids, embeddings, metadatas, documents = self._validate_embedding_set(

  File "/home/adminuser/venv/lib/python3.10/site-packages/chromadb/api/models/Collection.py", line 350, in _validate_embedding_set

    ids = validate_ids(maybe_cast_one_to_many(ids))

  File "/home/adminuser/venv/lib/python3.10/site-packages/chromadb/api/types.py", line 99, in maybe_cast_one_to_many

    if isinstance(target[0], (int, float)):

IndexError: list index out of range

Full code in my GitHub repo.

Thanks all!

[dosubot[bot]]

🤖

Hello,

From the error message you provided, it seems like the issue is not with the UnstructuredImageLoader but rather with the Chroma.from_documents method. The error message IndexError: list index out of range is thrown when you're trying to access an index that doesn't exist in the list.

In your case, it seems like the document_chunks list that you're passing to Chroma.from_documents might be empty. This could be due to the UnstructuredImageLoader not returning any documents when loading the image file, which would result in an empty documents list and subsequently an empty document_chunks list.

To debug this issue, I would suggest adding some print statements to check the contents of the documents and document_chunks lists after they are created. For example:

image_loader = UnstructuredImageLoader(file_path, mode="single")
image_data = image_loader.load()[0]
documents.append(image_data)
print(documents)  # Add this line

text_splitter = RecursiveCharacterTextSplitter(chunk_size=1500, chunk_overlap=150)
document_chunks = text_splitter.split_documents(documents)
print(document_chunks)  # And this line

embeddings = OpenAIEmbeddings()
vectorstore = Chroma.from_documents(document_chunks, embeddings)

This will help you verify whether the UnstructuredImageLoader is correctly loading the image file and whether the RecursiveCharacterTextSplitter is correctly splitting the documents.

If the documents list is empty, it means that the UnstructuredImageLoader is not correctly loading the image file. In this case, you might want to check whether the file_path is correct and whether the image file is in a format that the UnstructuredImageLoader can handle.

If the document_chunks list is empty, it means that the RecursiveCharacterTextSplitter is not correctly splitting the documents. In this case, you might want to check whether the chunk_size and chunk_overlap parameters are appropriate for your documents.

I hope this helps! Let me know if you have any other questions.

Sources

• libs/langchain/langchain/document_loaders/image.py
• libs/langchain/langchain/document_loaders/unstructured.py

---

This response is meant to be useful and save you time. It is not meant to be a precise solution, but rather a starting point for your own research.

Help me be more useful! Please leave a 👍 if this is helpful and 👎 if it is irrelevant.

If you want to continue the conversation, start your reply with @dosu-bot.

[nhtkid]

I misunderstood the usage of loader.

The UnstructuredImageLoader is used for OCR.
When your image has text in it, it can extract them out and there will not be any "List out of range" error.

I am getting that error because the image had no text in it.
My code is working fine.

To get to what I wanted, the loader I should use is the Salesforce ImageCaptionLoader. It will provide you a description of the image.