PLANET RECOGNITION AND CHATBOT

• 

  Planet Identifier

• 

  Astro CHATBOT

Inspiration

We were inspired to create this project because we wanted to challenge ourselves to build new things and try and build an image identifier.

What it does

The first part of the project is a Planet Detector which identifies the eight planets of the Solar System and the dwarf planet pluto. The second part is a ChatBot which will keep you entertained whenever you are bored.

How we built it

The project uses a picture as an input and detects which planet it is, using Machine Learning. The Planet Recognition system consists of a pertained ResNet50 Model with its last layer replaced by a softmax unit of 10 units. The model is Built using the Keras library. The weights of the ResNet50 are set to non-trainable while the weights of the softmax unit are trained using a small image dataset of planets downloaded from google images.

There is also a python chatbot which can crack jokes, greet you etc. We built the GUI using Tkinter library of Python. The ChatBot is built using NLTK which is a Natural Language Toolkit in Python. It is a rule-based chatbot that replies to simple questions like: ”What is your name?”, “Tell me a joke”, “Tell me a fact”. It is able to reply to simple questions by the user.

Challenges we ran into

The code was not working and we had to find a way to link the windows to each other. We both had to learn the syntax of Tkinter to be able to publish it.

What's next for PLANET RECOGNITION AND CHATBOT

We plan to make the chatbot better and add more features to it so works more efficiently. We also plan to train the Convolutional Neural Network on a bigger planet dataset so that it is more accurate in its predictions.

Built With

• machine-learning
• python
• tensor
• tkinter