ZooWind

Inspiration

Many NGOs have done research and investigations into the illegal wildlife trade (IWT) or engaged in counter-wildlife trafficking (CWT) activities. However, due to the lack of a centralized hub or a network organization, there has not been an effective tool for sharing all the available findings, tools, and materials among the conservation NGOs in Vietnam, resulting in waste of both human and material resources, and risks of losing the community’s trust because of overlapping of activities.

To solve the problem, our team has developed a platform to unite animal conservation NGO s ' resources using machine learning and deep learning.

What it does

Our platform has 3 main features.

1, Collect and share the resources from organizations, e-newspapers,... which classify and generated by our AI model.

2, Search and explore the animal information which can search by text or image which used deep learning to identify.

3, Report wildlife trafficking to authorities with our chatbot

How we built it

With the prototype we have designed on Figma, We built our platform using Angular 11, consuming RESTful API developed with Java Spring Boot.

In the document classification model, first, we collect over 4000 e-newspapers on the internet and label those using keyword extraction. After labeling, we built a model to classify text from the document using the Logistic Regression function into 3 categories: Not related, Illegal Wildlife Trafficking, Counter Wildlife Trafficking. The model precision is 84%.

In the animal classification model, we collect over 17k images from 5 species and put them into a deep learning model using CNN to train. The final precision is 67%.

After building 2 models, we integrate them to our front-end using Flask RESTful API.

Challenges we ran into

Our document classification is still not collecting all the results we expected.

Our website functions are not yet delivered all the needs of NGOs' problems.

Our image classification model accuracy is still low with small numbers of species.

Accomplishments that we're proud of

We came together with no relationship to each other, and finally, we worked well to create a product without any conflicts.

The product that we make is useful for not only researchers but also people who love wildlife

What we learned

We learned more about the current situation of wildlife trafficking, learned more about how to protect it, and make a product that we have ever had a chance to do. Also, we forced ourselves to learn something new, and have to use it for production in a short amount of time.

What's next for ZooWind

We will develop a platform for NGOs' including UI, API to help them upload and control the research, documents, papers that they want to deliver.

Built With

• angular.js
• cloud
• deep-learning
• heroku
• java
• machine-learning
• mongodb
• python
• spring