todo.md

SRC

main_stem.py

get_text_overlay()

1. Need to find a way to get PyTesseract to install automatically through a bash file and not have to go through the installation loop
2. Keras_ocr is also a good alternative, will need to check on it later on

get_overlay_box()

1. Need to find a good way to understand what threshold is good, might need to go a little more DL/CNN route
2. "--psm 6" feels right for now, can try the other ones, namely 11, 12 to see what else I can improve on the reading
3. Need to add black and white filter and see output differential
   1. Need to add thresholding to the filter applied
   2. Need to add output cleaning using re
4. Need to improve quality of text being read

object_detection()

1. Alternatives for object detection: yoloX, mmDetection, yoloV(4/7/8), openCV, moondream, timm

get_logo()

1. Make a function to get the logo from the image
   1. Can use logoHunter

Misc

Changes made in code

format: file_name - line of original code

1. vision_encoder.py - line 100
2. vision_encoder.py - line 40

Code that works

1. get_color_palette() this gets the color palette from the image and returns the top 5 most
2. get_overlay_box() this turns the image to black and white and then ocr's the image
3. object_detection() this uses mmDetection to detect objects in the image and returns the image with boxes around the objects