ImageToolbox
ImageToolbox copied to clipboard
T8RIN
Images analysis to search local images by text (pretrained static models)
I just came up with an idea that might be possible to implement but it's definitely one feature that could be considered as overkill but please hear me out. There is a tool called tidy which can be used to find images based on the description (Text-to-Image, locally quantized state-of-the-art vision-language pretrained CLIP model and ONNX Runtime inference engine). It uses models to map one ore more images to word descriptions. So it's more or less heavy.
https://github.com/slavabarkov/tidy
The use case for image toolbox might be that users can search within the app for a specific image and then immediately edit it. Another use case is that users search for a more or less specific expression of an image and then use it for example for artistic purposes like "show me all images with flowers" and then they can use it for example to make image collages and stuff like that.
Image toolbox has many functions needed already in place for example an image browser and other tools. It would be really great if you consider it as a feature in the long term.
I'm saying this because right now the development of tidy appears to be stalled for quite some time. Tidy is also written in Kotlin and works well.
@T8RIN It's definitely an impressive app. However, the functionality is already there in your app from my point of view. It's already possible to use a normal file manager that comes with image toolbox or the system to collect images. But there is no means to do this by machine learning methods (Text to Image), as in tidy. Therefore, I assume this would just consume a lot of your precious developer time and there are maybe more important things to implement. But thanks for considering it.
@T8RIN When I talked about the feature suggestion some time ago I totally misunderstood the capabilities of Gallery because I used an older version. The app was basically just a normal gallery app. Now I see what you mean. The app uses currently CLIP and they are going to extend it. Yes indeed it would be exactly what I meant.