Creating a deep learning pipeline for the identification of the person by the manner of its walking i.e. using his/her gait features.
The dataset that we will be using in the project will be the Human3.6M dataset. The dataset consists of 3.6 million different human poses collected with 4 digital cameras.
- Identifying and creating the Ground Truth data.
- Getting the individual poses for each of different concerned objects in each frame.(DensePoseRCNN)
- Establishing the spatio-temporal relationships between the frames in the gait cycle using RNNs or LSTMs.
- Optimizing the above network by creating TensorRT engine to work on the Nvidia Jetson .Tx2
The Assignment's aim is to develop human recognition system via gait features. This Assignment was under Prof. Aditya Nigam.
- Github: http://github.com/abhijeet2096
- Email: sharma.abhijeet2096@gmail.com
- Mobile: +91-8629015433
- Github: https://github.com/mohitsharma1996
- Email: mohit21sharma.ms@gmail.com
- Mobile: +91-8629015362
- Github: https://github.com/akhilsinghal1234
- Email: akhilsinghal1234@gmail.com
- Mobile: +91-8629015410