A simple Keras model to detect whether a video has happiness inducing contents or not.
- These days we get lots of forwarded WhatsApp videos and there contents can very upto wide range which can make us feel dejected for no reason.
My motive behind creating this model is to know in advance whether a video has happiness inducing factors or not.
- Keras
- Opencv '3.2.0'
- ffmpeg '3.2.4'
- Make sure they both are compatible with each other otherwise it's 100% painful to struggle with Videocapture.
- Model for now just detects happiness in the face captured inside a frame. Moreover, I have implemented the model on a very small dataset with 600 training images and about 150 test images. We can surely increase the accuracy with more precise dataset.
- This project for now lies in it's very early stage, we can surely implement speech recognition (may be some NLP as well), picture perception test, object detection and some text segmentation to make this model bullet proof.
- We can also make an Android Application out of this model as well where we just need to open the video with our application and our application can tell whether one should see the video.
- Peace..!!