https://github.com/humblecoder612/SAR_yolov3

Best Accruacy:speed ratio SAR Ship detection in the world.
https://github.com/humblecoder612/SAR_yolov3

Last synced: 5 months ago
JSON representation

Best Accruacy:speed ratio SAR Ship detection in the world.

• Host: GitHub
• URL: https://github.com/humblecoder612/SAR_yolov3
• Owner: humblecoder612
• Created: 2020-05-02T12:53:40.000Z (almost 5 years ago)
• Default Branch: master
• Last Pushed: 2020-05-03T18:44:15.000Z (almost 5 years ago)
• Last Synced: 2024-08-02T01:24:11.824Z (9 months ago)
• Language: Python
• Homepage:
• Size: 34.2 KB
• Stars: 19
• Watchers: 2
• Forks: 2
• Open Issues: 0
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

• awesome-yolo-object-detection - humblecoder612/SAR_yolov3
• awesome-yolo-object-detection - humblecoder612/SAR_yolov3

README

        
# SAR_yolov3

# Welcome to SAR SHIP DETECTION

We have applied YOLO-V3 Object detection on  **SAR Satellite** Images. We are detecting ship by these Images where SAR sensors are immune to bad weather and night time which a great way of detecting . We applied YOLO-V3 to it and it gives best accuracy:speed ratio in the world among all other models and methods applied. We are further looking to improve the accuracy where the current accuracy is **90.25 %**.

# Files

We have included the Config file which is the model architecture for the **darknet** deep learning framework where we changed the models in many ways for our results , from data augmentation to hyper-parameters.

Anchors boxes is also given in this repositories.

Some scripts which converted the **VOC Xml**  into **Darknet** Text Format

## Results

![Prediction 1](https://github.com/humblecoder612/SAR_yolov3/blob/master/images/1.jpg)

**PREDICTION 1**

![Prediction 2](https://github.com/humblecoder612/SAR_yolov3/blob/master/images/2.jpg)

**PREDICTION 1**

## Submission 

We have written a research paper on this project and submitted into a Springer conference.