Description: Translate images to emoticons 👏😌
- OpenCV2
- NumPy
- Scipy
- Sklearn
- DepthAI (optional - only if you have a DepthAI camera)
- Clone this GitHub project:
git clone
- Install requirements:
pip install -r requirements.txt
- Download Twitter Emoji library here, extract the
72x72
folder under./assets/
, and rename folder to "twitter". - Run script:
- Calculating from scratch:
python main.py
(NOTE: Calculating cache will take ~1.5 hours ) - Calculating from cache is just the same command. You can get the cache here
- Calculating from scratch:
Performance: On my machine (AMD Ryzen 2700X), I was able to maintain 38-42 FPS. If using a lower-spec CPU (this is pretty CPU dependent), then try:
- Decreasing video size
- Increasing pixelization size
usage: main.py [-h] [--video VIDEO] [--output OUTPUT] [--depthai]
optional arguments:
-h, --help show this help message and exit
--video VIDEO Video input. Enter either a number for a webcam device or
path to a video file. Example: python main.py --video
test.mp4
--output OUTPUT Video output path. Enter a string to the path where the
video output will be saved. Default is a .mp4 file.
Example: python main.py --output out.mp4
--depthai Optional flag only used for DepthAI devices. Flag is false
by default. After enabling program will import depthai
module.
Optimize algorithm, rethink mapping algorithm, and program as some kind of filter for a social media platform. Hmmmm.... could multithreading help speed things up?