person_detection_logistic_regression

Project to detect pedestrian from a video using logistic regression with neural networks

Prerequisites

you have to have previously installed

Python 3.7

Pip 3

Requierements

Virtual Enviroment

Installation

git clone https://github.com/DavidBanda/person_detection_logistic_regression

python3 -m venv person_detection_logistic_regression

person_detection_logistic_regression\Scripts\activate.bat

cd person_detection_logistic_regression

pip3 install -r requirements.txt

Run project

python run.py

You can access it opening in your browser index.html

You need to write down in the text field an absolute path to a video to your local machine, you can use the videos that are in the project: person_detection_logistic_regression/ai_backend/static/video

Create your own dataset to detect objects/animals/persons

Image dataset

You need images of the object you want to detect and images in which the object is not. You can use this site Kaggle to download datasets.

Rename Image dataset

Once you download images, you need to tag the images with their respective name, for example we tag our images with the name 'person' next to a number and 'obj' next to a number for images that do not appear a pedestrian. You can check it in the project at the route person_detection_logistic_regression/ai_backend/static/images/object_dataset

To facilitate this, we create a file to rename all files in a folder, the file its in person_detection_logistic_regression/ai_backend/utils/rename_files.py, inside we only need to change the variable path to the path where we have the images that we want to change the name. We also need change the variable name for the name that we will use plus the number we gonna add to each image, but this number is adding automatically.

Once you labeled the images you need to move them to person_detection_logistic_regression/ai_backend/static/images/object_dataset

Create HDF5 file

We need to create a HDF5 file to train our neural network. To do this we need to go at the file in person_detection_logistic_regression/ai_backend/utils/dataset_h5_creator.py and run it. But first we need to update the name variable with the name we use to label our image object dataset, in our case was person.

Run project

Finally just run the project.

python run.py