Skip to content

The goal of this competition is to accurately identify starfish in real-time by building an object detection model trained on underwater videos of coral reefs.

Notifications You must be signed in to change notification settings

lolloloschi97/tensorflow-great-barrer-reef

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

51 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Help Protect the Great Barrier Reef

Abstract

The project is about object detection of crown-of-thorns starfish in underwater image data. The reference Dataset has been taken from the Kaggle Challenge.

Implementation

After properly preprocessing the dataset, we have exploited RetinaNet architecture with a ResNet50-FPN as backbone, pretrained on coco 2017. In order to obtain a coherent prediction, we performed a fine-tuning on our data of the last stage of the backbone architecture and the remaining RetinaNet structure (Anchor generator, classification head and detection head). We trained for 10 epochs with batch size equal to 4 and a learning rate equal to 1e-5. During training we also added as a regularization technique a random horizontal flip with probability 0.5.

Results

We have obtained good results, reaching an Average Precision of almost 0.9 in the validation set. We also tried to test our model with unseen images downloaded from the web to understand in a better way its performances.



We can observe that not all crown-of-thorns starfish are correctly recognized and also in some cases, the bounding boxes were not completely covering all the starfish but just a portion of them. Here there are some examples:

Contribution

About

The goal of this competition is to accurately identify starfish in real-time by building an object detection model trained on underwater videos of coral reefs.

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages