Object-Detection-Model

The following machine learning model aims to identify objects in the video and predict the trajecory that they are moving in. The image processing algorithm will detect instances of semantic objects of a certain class (such as humans, buildings, or cars) in digital images and videos.

Deployment Instructions

Ensure that you have the following for both GPU and CPU Installations:

• PyCharm Community Edition or any other IDE of your choice
• Anaconda (https://www.anaconda.com/products/distribution)
• NVIDIA Drivers (https://www.nvidia.com/Download/index.aspx)
• CUDA Toolkit (https://developer.nvidia.com/cuda-downloads)

Model Performance on Mall CCTV Camera

Model Performance on Figure Skaters

Model Performance on Highway Traffic

Model Performance on Soccer Players

Model Performance on a Tennis Player

GPU Installation, set-up a virtual environment with the following command

  conda env create -f environment.yml

Then activate your environment

  conda activate <Name_of_the_Project>

Check which drivers you are using

  nvcc --version

You should be seeing the following

  nvcc: NVIDIA (R) Cuda compiler driver
  Copyright (c) 2005-2019 NVIDIA Corporation
  Built on Fri_Feb__8_19:08:26_Pacific_Standard_Time_2019
  Cuda compilation tools, release 10.1, V10.1.105

Make sure conda is saved in your Environment Variables (PATH)

   C:\Users\Daniil_Zhilyayev\Anaconda3\Scripts

   C:\Users\Daniil_Zhilyayev\Anaconda3

   C:\Users\Daniil_Zhilyayev\Anaconda3\Library\bin

You are all set, here is a few commands for you to get started:

  python motioned_detection.py --source videos/yolor/F1_CARS_DETECTION.mp4 --cfg cfg/yolor_p6.cfg --weights yolor_p6.pt --conf 0.25 --img-size 1280 --device 0 --view-img

  python motioned_detection.py --source videos/yolor/F2_MALL_DETECTION.mp4 --cfg cfg/yolor_p6.cfg --weights yolor_p6.pt --conf 0.50 --img-size 1280 --device 0 --view-img

  python motioned_detection.py --source videos/yolor/F5_SKATING_DETECTION.mp4 --cfg cfg/yolor_p6.cfg --weights yolor_p6.pt --conf 0.75 --img-size 1280 --device 0 --view-img