Jhony32

To defeat the Wildlife Trafficking, we proposed a platform of denounce with options of type, photo and locations of the crime. Besides that, with the help of ML we can identify the reported animals.

Comment

Inspiration

The Wildlife Traffic is the 3º biggest traffic in the word. The consequences of this type of traffic is the extinction of animals affecting the nature and the world. In Brazil the government agents has a huge difficult in defeat the wildlife traffic, one of the problems is the quality of the denouncements. Most of the cases the denouncements is made without important data like locations, what animal is or the type of crime. If the identify was quick the agents could avoid the animals death after the capture, this is a problem because in consequence of the big Brazilian fauna, the agents has difficulty in identify and take care on the right way of the animals.

What it does

Jhony32, is a platform that enable their users do report Wildlife Traffic crimes. The report is made through occurrence classifications (TAG's), photos, GPS location and some minor information. Each report sustain a DataBase which is then presented in a Power BI Dashboard to the responsible Organizations, such as IBAMA and PRF in Brazil, by ordering the most relevant occurrences to help these agents do prioritize and enhance their actions and decisions.

How we built it

The platform was based on the Kivy developed App, made in python. The App uses a Machine Learning Algorithm to recognize Animal species. In the developed case, it was trained to differenciate the 22 most trafficked birds species in Brazil. To create the model, it is utilized a Microsoft Azure tool, training it with almost 100 photos per bird specie. The model achieved a 95% Accuracy and a 86% Recall, with a threshold of 75%. It means that the model predicted 86% images with a class certainty over 75% and 95% of these predictions where correct.

What we learned

Wildlife Traffic is a serious and difficult problem to solve. The information is scarce and the reports have few information, which prevents the effective action of environmental agents. From a technical standpoint we were able to learn how to develop a solution integrated not only by the application users but also by the environmental agents themselves. We learned to connect all the solution steps using Python and Power BI. We also learned how to divide tasks according to each team component skills and how important it is to define the problem to be solved and everyone to be aligned in their specific roles.

What's next for Jhony32

The next steps for Jhony32 is to raise a solid DataBase, improving our Species Recognition Algorithm, with more generalist predictions, encompassing also sex, age and injuries differences. Nevertheless, our main goal is to build a strong network through Marketing actions, which will enhance our efforts to end Wildlife Traffic. We plan to improve our platform as an App and a Web Service to connect as many Wildlife lovers as possible and spread the platform, initially through Brazil, and then, through the World!

More Information

To see more information and details of our project please access the docs.google.com link in Try it out section. Also to see more details about the implementation see the GitHub link in the same section.

Built With

Share this project:

Updates