ImageAI VS video_stream_opencv

Hi. I want to implement an image server similar to Photoprism using ImageAI to tag images based on objects and context. However I don't want to spend to much time working on the frontend, at first I were thinking about using Danbooru and use Flexbooru or the web interface on my phone. But it doesn't have any encryption or password protection (since the purpose of it is to be used as a public image board).

From ImageAI?

https://github.com/amusi/awesome-object-detection https://mmdetection.readthedocs.io/en/latest/index.html https://github.com/thtrieu/darkflow https://github.com/OlafenwaMoses/ImageAI https://github.com/dmlc/gluon-cv https://github.com/aim-uofa/AdelaiDet/ https://github.com/aim-uofa/AdelaiDet/blob/master/configs/FCOS-Detection/README.md https://github.com/wizyoung/YOLOv3_TensorFlow

Some resources to get you started...https://towardsdatascience.com/object-detection-with-10-lines-of-code-d6cb4d86f606https://github.com/OlafenwaMoses/ImageAIhttps://towardsdatascience.com/yolo-object-detection-with-opencv-and-python-21e50ac599e9https://github.com/meituan/YOLOv6

Thanks for your suggestion, I appreciate it! I think you're right. I stumbled upon the ImageAI project which I think might make for good interface to train the CNN. Going to give that a try and see how the model turns out.

Image AI

Yo good question, my first thought was image recognition so something like this (https://github.com/OlafenwaMoses/ImageAI) and and then some other stuff so it is able to find the search bar, click on it, and then input the text. Or you could use image recognition to copy the URL and then open up a Selenium instance from there.

Use a pre-trained model such as YOLOv3 to feed datasets into for Image recognition so model can learn what you want then instead of using Opencv to run model for image recognition you could use ImageAi

It's the default launch file and it seemed pretty good (video0, 30fps..)

Video Streaming