diff --git a/Deep Fusion AI/Subhasmita-Swain.md b/Deep Fusion AI/Subhasmita-Swain.md new file mode 100644 index 0000000..2571a92 --- /dev/null +++ b/Deep Fusion AI/Subhasmita-Swain.md @@ -0,0 +1,29 @@ +# Developer - Winter of Code 2020 +# Subhasmita Swain +### Deep Fusion AI : Social Distance Detector + +This project aimed to build a light weight and accurate mobile friendly Social Distance Detection model. My work throughout this month was to produce a viable python script for detection (using avalable Tensorflow Models) and enumerate a distance measurement algorithm between pedestrians. + +I am [Subhasmita Swain](https://github.com/SubhasmitaSw) an undergraduate from Sharda University, as a part of Winter of Code. + +This project was mentored by [Sayak Paul](https://github.com/sayakpaul) and [Rishiraj Acharya](https://github.com/rishiraj). + +# Contributions + Here's my [colab](https://colab.research.google.com/drive/1ZSdQ3eVCBLejNgE-f96fHHv1eEohlp7s?usp=sharing) notebook containing my work. +- Tested different pre-trained TF models for their performance compyling to the Detector +- Prepared detection and distance measurement algorithm scripts using ssd-mobilenet-v2 model +- Tested the performance of TinyYOLO on NVIDIA Jetson Nano V2 +- Provided Documentation for the final project. + + +# Issue +- Here's the [Issue](https://github.com/DeepFusionAI/social-distance-detector/issues/6#issuecomment-751362640), I've co-worked on. + + +# Future Scope +With the upcoming support of TFLite for the EfficientDet model can help improve the accuracy and speed of the Social Distance Detector with more mobile-friendly models. +Also with further Optimization of the model we could push the perfromace bar ahead a little. + +# Overall Experience +The constant support and relentless doubt clearings from my mentors Sayak Paul and Rishiraj Acharya helped develop my technical side as well gave me an insight about professional work culture. +I would also like to Thank my teammates Piysuh, Sudarshana and Hiran, I've learnt a lot from you.