Extracting text from video with pytesseract.

First clone the repository.

git clone https://github.com/saty101/image-to-text.git
cd image-to-text/
pip install -r requirements.txt

If you encounter file note found as below:

FileNotFoundError: [Errno 2] No such file or directory: 'tesseract'

Run the following command -

apt install tesseract

You can download a youtube video using the youtube_video.py and change the link to your desired video.

python youtube_video.py

Since we mainly work with news videos/videos with text embedded in it, it will be saved as news.mp4 in the same directory. The mp4 file is downloaded in the same directory.

For annotating the videos, run

python main.py

This will create a labeled_news.mp4 along with annotations.txt where you can find the annotated video along with a file of captions produced in the video.

You can also find the kaggle notebook where you can find the code with embedded videos(along with the results)

The full workflow code is in this notebook which shows how annotations with one image and then creating the whole vide0 but the videos are not visible in the jupyter notebook. Colab notebook link of the same notebook - link

Simply change the link = <desired link> to annotate the video you want to.

Colab noteboook but the videos can only be downloaded as embedded videos in colab crash and cause the runtime to start again.