Support images / binary input in Studio

[SergioRubio01]

Hey everyone:

During my latest project, I came across the idea of introducing a vision language to process images. Some parts of my code can be seen below for a two-agent collaboration workflow.

def process_single_image() -> dict:
    """Process a single image and return the results."""
    
    # Only use this for langgraph studio debugging --------------------------------------
    image_directory = 'C:/Users/Images/'
    image_path = [os.listdir(image_directory)[-1]]
    # ----------------------------------------------------------------------------------
    try:
        # Encode the image
        base64_image = encode_image(image_path)
        
        # Create initial state
        initial_state = {
            "messages": [
                HumanMessage(content=[{
                    "type": "image_url",
                    "image_url": {
                        "url": f"data:image/png;base64,{base64_image}"
                    }
                }])
            ],
            "sender": "human",
            "file_path": image_path
        }
        
        # Run the graph
        result = graph.invoke(initial_state)
        return result
        
    except Exception as e:
        print(f"Error processing {image_path}: {str(e)}")
        return {"error": str(e)}

How can I import an image directly in the Langgraph Studio platform? Maybe it is impossible yet if someone else knows better, please let me know :)

[dqbd]

Hello! Image / binary input is not supported yet in Studio