A binary classifier to detect the presence of human in an image. A simple convolutional network has been applied with one hidden layer and L2 regularisation. Data augmentation has been done by the ImageDataGenerator function of keras. Training set of around 7000 images has been used and the results are been validated after each epoch with around 800 images.
windows 7
python: 3, Jupyter notebook
libraries: Numpy, Keras
- download the code and open it in jupyter notebook
- install necessary libraries
- give appropriate address to the training and test data ( The datas can be downloaded from the link provided)
- Train the model
- Specify the test image address for testing on a particular image
link for data -(https://drive.google.com/file/d/1pLupzm09JKqKJ1AJ_XUfKKllzlZSDoeZ/view?usp=sharing)