svp19 / plithos

Crowd analysis using Tensorflow's pre-trained ssd_mobilenet_coco_2017

People Detection using TensorFlow

The program aims at seeing how useful the pre-trained model from tensorflow : ssd_mobilenet_v1_coco_2017_11_17 can be to count the number of people in a variety of places ranging from Public restaurants to crowded market places.

Results for some simple images shown in test_images. This project is still a work in progress and more info may be found in the below section.

Future Scope

Transfer learning may be performed to improve the head count in crowded places and the model may be used to estimate the growth rate of number of people in crowded places which is an essential parameter in crowd control analysis. The model may also be deployed for queue analysis to estimate waiting times for users.