copyright

only mention sources from major leagues (#fans > 5k)

mention the bgm you are using

non-max suppression in combining similar bounding boxes

the lib:

1
2

from imutils.object_detection import non_max_suppression

basically greek letters

maybe you can document another great range of symbols by just enabling the system to search in greek?

could also search among math symbols, do math ocr.

kindly reminders

when building python c++ libraries without xcode, please add commandline header files like this:

in order to have this during build:

/Library/Developer/CommandLineTools/Library/Frameworks/Python3.framework/Versions/3.8/Headers/Python.h
we need to do:

1
2
3

ln -s /Library/Developer/CommandLineTools /Applications/Xcode.app/Contents/Developer

search by image instead of cranking latex out

image search libraries

character level optical char segmentation called chargrid ocr
image match used for copyright violation detection, using phash algorithm

get the latex out

first detect the location of math formula

scanning single shot detection for math formulas
dataset for math symbol detection
detect different part of document with yolov3
math expression detection

next find the tool for picture to latex conversion

新开源的Python工具——Pix2Text (P2T)，目标是 Mathpix 的Python开源替代品，现在可以识别截图中的数学公式并转换为Latex表示，也可以识别图片中的中英文文字。在线Demo： https://huggingface.co/spaces/breezedeus/pix2text 。 Github: https://github.com/breezedeus/pix2text ，Gitee: https://gitee.com/breezedeus/pix2text 。
attention based math ocr
gui of image2latex
im2latex-tensorflow
im2markup with only nvidia support using torch, model for latex conversion can be found here
deeplearning picture to latex
pix2tex
using pix2tex
中文公式 手写公式识别 需要进一步训练
中文 手写 pytorch版本

mask latex area and get conventional things out

easyocr with pytorch support

search the formula

formula search based on sympy
can we render latex to picture with sympy?
latex search engine