Introducing WeedCam
Mobile application that recognizes
how stoned a person looks in a photo
Welcome to WeedCam - mobile application that recognizes how stoned a person is by photo.
Application analyzes the face and taking into account eyes width, eyes redness, eyelid position, visage color and other features provides you with an overall marijuana effect. You can take a new photo or use previously saved from your gallery.
WeedCam can help you to understand what you smoke, how strong the kush strain is. THC, sativa indica level, and weed terpenes of cannabis affect your look too. Also, using to our application, you can understand whether you should drive, go to work, or is it better to wait a bit, or maybe even take a day off.
From 420 enthusiasts to medical marijuana advocates, WeedCam can be useful for all kinds of stoners. No matter what you like to smoke: bong, blunt, joint, or different kinds of pipes, no matter how you call it: marijuana, cannabis, ganja, kush or may be pot, dope, hemp or herb - our weed app will be helpful for you.
An extended research, devoted to eyes and face recognition problem was conducted. Final algorithm consists of next steps:
This step is performed to achieve picture segmentation and determine which part of the photo will be analyzed.
In this step, we apply an eye detection algorithm. We measure the eye parameters and calculate the width-to-height ratio.
Using the eye positions obtained in the previous step, we translate the image into different color spaces and then calculate the eye redness value from specific channels.
Using the face segmentation from the first step, we translate the image to different color spaces and then calculate the visage color value.
In the final step, we use the model pre-trained on partial features to calculate the total effect value.
To train our model, we gathered a dataset of several thousand photos. The first part of the database consists of photos of various people from the internet to enhance the algorithm's generalization ability. The second part includes photos from live experiments conducted in wild.