BOTW

https://youtu.be/JPr98zvZuKQ

https://botw.life/

In this project we developed the object detection model (YOLO) to detect garbage on the waterways, now we are able to recognize 93.6% of garbage on out test set.

Initially, the very beginning of all Machine Learning models is to prepare the dataset, this is very predominant as good data always leads to a better outcome. Thankfully, we have a tremendous dataset collected from the main canal, with the help of friends, locals and volunteers all around the country.

we use "LINE" as a social network which famous in Thailand to collect garbage images from volunteers. We use Roboflow as a tool to mobilize volunteers to do image annotation.

As we have good, collected dataset, it is even developed with our ingenuity augmentation, the dataset is immensely emerging, from thousands to ten thousand images, with all types of predicted scenario that the scene would endured.

Consequently, we had trained YoloV5 for 100 Epoch when we get the desired model.

In further develop it can be both trash classification and waste tracking in the canal to create a trash surveillance system integrated with the government in the future.

When installing a smart surveillance camera Entering various sub-line, our model is able to track waste direction, classification and notify them as schedule.

to the staff and people along the canal with volunteer spirit can come to help manage such waste groups.

 In addition to this process, the state has shifted its view of disruption to “proactive”. It also allows us to start screening zones of residential neighborhoods that are at risk of dumping waste into the canals.

We saw NASA's open sources and Our's project can help Debris Tracker to take less human involved and many others.

We hope to automat the way we track marine debris

In this year of 2021, we have high initiative for this SpaceApp Hackathon with the wide - range of skill capabilities bringing in the profession Artificial Intelligence and Web Application from Thailand.

As a team, we came together for this fascinating name, BOTW abbreviates for Bed On The Water; this rises into the question of “Are we now in the state that all the waste, no matter what size and shape it is could be found in the canal?”, this canal was once clean enough to take a bath and swim in it, and the abundant trash dump is totally covered the surface of water, damaging the eco-system, making stinky smells and decaying the life and even human habits, almost like the apocalypse of the environment. We have to ACT! This is not okay! Obviously, we can make this habitat better and cleaning up these putrid streams of water. We will make it together!

Throughout this Hackathon, it has emphasizes the impact of , it has deepen our knowledge and the recognition of water debris. It is now our task to “Leverage” this Machine Learning to solve the profuse negative undertone of waste. Initially, the very beginning of all Machine Learning models is to prepare the dataset, this is very predominant as good data always leads to a better outcome. Thankfully, we have a tremendous dataset collected from the main canal, with the help of friends, locals and volunteers all around the country. As we have good collected dataset, it is even developed with our ingenuity augmentation, the dataset is immensely emerging with all types of predicted scenario that the scene would endured.

YoloV5 Weight https://drive.google.com/file/u/1/d/1RrdhBZCZaXsdCCjgwNYxf7xZdNspJ816/view?usp=sharing \n Dashboard https://wandb.ai/colabstudy/YOLOv5/runs/19edp2nh?workspace=user-colabstudy

#yolo, #canal, #debris, #detection ,#bed ,#bed_on_the_water #botw ,#bubble_barrier, #water

This project has been submitted for consideration during the Judging process.