Kinesics

KINESICS
Happy Face
Sad Face
Love
Victorious

Inspiration

The most important part of communication is hearing what isn't said. Our body language tells a lot about us. When people are happy they smile ;when they are sad or angry they frown or scowl. Ascribing emotions or intent to body language and facial expressions is literally in our genes. Most of the basic communication gestures are the same all over the world. On the other hand, just as a verbal language differs from culture, so the non-verbal language may also differ. one gesture may be common in a particular culture and have a clear interpretation, it may be meaningless in another culture or even have a completely opposite meaning.

Learning to decode nonverbal communication can help us understand other people’s thoughts and feelings, and it can help us make ourselves better understood too. Whether it’s friends, family or lovers, colleagues or strangers, there is a wealth of meaning in the body language of those around us and most of us aren’t even aware it exists – Body Language Decoder can help you tap into it.

What it does

It detects the facial features, and body joints of a person and based on the person's behavior it determines their current emotion.

How we built it

Basically, we've leveraged Mediapipe to estimate both facial and body landmarks. With that data we were able to build custom pose classification models which allows us to decode what a person might be saying with their body language.

Challenges we ran into

Getting the predicted emotions to be as accurate as possible was a bit challenging.

Accomplishments that we're proud of

We're successfully able to detect facial outlines, body pose and predict the emotions. We've made this in one night which is a huge thing for us as a beginner.

What we learned

We learned how well we can customize our body language with the help of features given by Mediapipe line.

What's next for Kinesics

We are looking forward to develop a full-fledged website ,where our Kinesics can be further extended to assist people prepare for job interviews. The idea is to analyze body posture and give some feedback on it over a given period of time.

Built With

mediapipe
opencv
python

Submitted to

Scioverflow(2022)

Created by

In this project, I've contributed to the understanding and implementation of the pose estimation system with Mediapipe. Using
the holistic Media Pipe library I compared the gestures captured earlier with the predefined body_language.pkl file. Then it predicts the gesture.
Also, it predicts the accuracy of the pose compared to pre-trained gestures.

Tejaswi kottakki
In this project, I've worked on the OpenCV and on how to perform training using pose estimation systems and analyze the applicability of the estimation systems to body language recognition. I even worked on the integration and storing of various coordinates of facial expressions and body gestures captured and exported to a CSV file named coords.csv.

PERI VISHWANADHA
In this project, I have worked on understanding the MediaPipe Body Landmark model which basically gives for high-fidelity body pose tracking,inferring 33 2D landmarks on the body from RGB video frames, utilizing BlazePose. I have even worked on the concept of how emotion detection from facial expressions is possible and helped in implementing it in the training part of our machine learning model.

Shreya Chintawar .
Harika Nethi

Updates

Tejaswi kottakki started this project — Aug 14, 2022 04:41 AM EDT

Leave feedback in the comments!

Log in or sign up for Devpost to join the conversation.