ReptilePredictor is a web application that can predict if a reptile image uploaded by the user shows a snake, a lizard or a crocodile. The backend of this web app is a transfer learning ResNet50 model implemented in Keras. The frontend, on the other hand, was made using Dash.
To run the website locally, please perform the following steps.
- Clone this repository into a folder in your machine
git clone https://github.com/meraldoantonio/ReptilePredictor.git
cd ReptilePredictor
- Create a virtual environment and install the dependencies within the virtual environment
conda create -n reptilepredictorenv python=3.6
conda activate reptilepredictorenv
pip install -r requirements.txt
- Run the application. The application will download and serve the model.
python main.py
-
Upload a jpeg image of either a snake, a crocodile or a lizard in the field provided. Your image has to be in jpeg format!
-
The probability assigned to each class will be shown as a bar chart.
-
Hover over the bar chart to load additional information about the probability of each class.