AI Image File Caption Utility скачать в хорошем качестве

AI Image File Caption Utility

In this project I use Langchain and AI to update the file names of image files. For my use; I also convert .heic files to jpg, and add some conditions for when .mov files should get copied. Some specific points of interest for me were; how to include image files in Langchain prompts, how to get structured data responses from an AI model, how to work with .heic files and use argparse to define command line arguments. 00:04 Introduction 01:14 Look at iOS photo album files 03:39 Image meta-data 04:30 Walk through the process flow I will use for my Langchain/AI program. 07:23 Review the python libraries I use in the project. 10:03 I use open-webui to test different Ollama hosted vision models. I decided on using granite3.2-vision:2b because of its small size, speed and still gives good feedback on the content of images. 12:34 I walk though the aiImageCaption.py code. 14:13 I use pydantic to define the data structure that the model will use to format the output from prompts. I use this to make sure it gives me 4 keywords for the image, in a format I can use. Later I will show how I use this when invoking the AI vision model. 14:33 I load a .env file that contains parameters used to perform Langsmith traces. See Langsmith link below to see more about how to use Langsmith for Langchain instrumentation. 15:26 I set up my argparse settings. This helps with creating a command line application that will accepts specific arguments. For my use I expect a source and destination folder for the images. Optionally I can specify a model and base_url for my Ollama instance. 17:20 How I use pillow-heif to convert heic files to jpg. 18:53: The main processing loop tat drives the application though the image files and folders in the source directory. 25:52 Demonstration of running the AI Image File Caption application. 28:06 I used pyinstaller to build a windows exe to make running the utility easier in the future. 28:45 My github repository for this code. It includes the full project including the “uv” package manager files. I also include the dist folder that holds the aiImageCaption.exe version of the application. References Github repository: https://github.com/somervda/aiImageCa... Langchain: https://docs.langchain.com/oss/python... Langchain Structured Output: https://docs.langchain.com/oss/javasc... Langchain multimodal messages: https://docs.langchain.com/oss/python... argparse: https://docs.python.org/3/library/arg... Langsmith: https://docs.langchain.com/langsmith/... Pillow-heif: https://pillow-heif.readthedocs.io/en... Ollama: https://ollama.com/ Ollama Vision Models: https://ollama.com/search?c=vision Open-Webui: https://github.com/open-webui/open-webui