Page Link: https://shan18.github.io/Flash/playground
This directory contains the code to setup the lambda modules present in the Playground section.
Pose Tracking is the task of estimating multi-person human poses in videos and assigning unique instance IDs for each keypoint across frames. Accurate estimation of human keypoint-trajectories is useful for human action recognition, human interaction understanding, motion capture and animation.
For code and instructions on how to setup the lambda function, checkout the human_pose_estimation directory.
| Input Image | Output Image |
|---|---|
 |
 |
Neural style transfer is an artificial system based on a Deep Neural Network that creates artistic images of high perceptual quality. The system uses neural representations to separate and recombine content and style of arbitrary images, providing a neural algorithm for the creation of artistic images.
For code and instructions on how to setup the lambda function, checkout the neural_style_transfer directory.
| Input Image | Style Image | Output Image |
|---|---|---|
 |
 |
 |
 |
 |
 |
- AWS Lambda
- Serverless
- AWS S3
- PyTorch
- Torchvision
The playground module contains 2 lambda functions
human_pose_estimation: Draws pose of a human by detecting keypoints in the input image.neural_style_transfer: Applies a style of choice to any given input image.
- Each lambda function has its own page in the playground section of the website.
- User goes to the respective lambda function's page and uploads the required input data.
- After clicking submit, the corresponding lambda function fetches the model from S3, performs inference and returns the output to the fronend.
- The frontend shows the returned output to the user on the webpage.