Missing Extracted Content with Crew AI Vision Tool

CrewAI Community Support
1

Description:
I’m using the Crew AI Vision tool for content extraction from images, but some content is consistently missing in the output. When I try using a custom tool with OpenAI, I receive the full content without any omissions. It seems there might be a discrepancy in content extraction accuracy between Crew AI Vision and OpenAI’s custom tool.

Steps to Reproduce:

  1. Use the Crew AI Vision tool to extract content from an image.
  2. Compare the output with a full extraction from OpenAI’s tool.
2

Has anyone been able to provide an example of how VisionTool can be used to communicate between agents or to perform a specific task of extracting text from what is seen in an image? The documentation is poor and there are many people commenting on problems and inconsistencies in the tool. A working example will help to clarify.