AI-Powered Sloth Speciation

Inspiration

The inspiration for this project originates from the need for accurate species identification. While distinguishing between two and three-toed sloths might seem straightforward, the natural world frequently presents us with far more complex challenges. Consider, for instance, the Amazonian river dolphin (Inia geoffrensis), a species with variations among sub-species that are not easily distinguishable through physical attributes alone. In this context, precise classification becomes paramount, especially in the context of rescue and rehabilitation, where a misidentification could have profound consequences. This project is a response to the demand for a dependable AI-based solution that could adeptly tackle the intricacies of species identification, ensuring that we can better care for the remarkable diversity of life that shares our planet!

What it does

This AI is designed to perform species identification for sloths based on their physical characteristics. It takes various measurable attributes such as claw length, size, tail length, and weight as input features. The AI analyzes these features and accurately predicts whether a sloth belongs to the two-toed or three-toed species.

How we built it

It starts by loading a dataset containing sloth data from a CSV file and then preprocesses this data, eliminating unnecessary columns and converting the 'specie' column into numerical values, making it suitable for binary classification. Next, the dataset is split into feature variables and the target variable, with 80% reserved for training and 20% for testing. To enhance model training, feature scaling is applied using StandardScaler. The neural network model, constructed using TensorFlow, includes an input layer with 256 neurons, a hidden layer with 256 neurons, and an output layer with a sigmoid activation function for binary classification. After compiling the model with the 'adam' optimizer and 'binary_crossentropy' loss function, it undergoes 50 epochs of training with a batch size of 32, with performance evaluation conducted on the test data.

Challenges we ran into

The conversion of the 'specie' column into numerical values to facilitate binary classification was the primary challenge that this AI encountered as the 'specie' column contained categorical values representing the two sloth species: 'two_toed' and 'three_toed.' To address this challenge, a mapping technique was employed, assigning 'two_toed' to the numeric value 1 and 'three_toed' to 2. While this conversion seemed straightforward, it demanded careful handling to ensure that the numerical values accurately represented the distinct sloth species. Any misalignment or incorrect mapping could have led to misclassification and inaccurate predictions by the AI.

Accomplishments that we're proud of

One of the accomplishments I take pride in is my approach to data preprocessing. In particular, I successfully addressed the need to prepare the sloth dataset for binary classification. This achievement involved a multi-step process that significantly improved the quality and suitability of the data for our AI model.

What we learned

I discovered that machine learning models are adaptable to various types of data. While our dataset focused on sloths, the skills acquired can be applied to other species and classification tasks.

What's next for AI-Powered Sloth Speciation

While this dataset was specifically tailored to sloths, I acknowledge that the skills I've acquired during this project are readily transferable to other species and classification challenges. Precise species identification serves as a linchpin in conservation endeavors, enabling targeted protection and management strategies. This AI model stands as a testament to the role technology can play in preserving biodiversity.

Built With

• jupyter
• pandas
• python
• scikit-learn
• standardscaler
• tensorflow