Airbus Ships detection problem

This a computer vision object detection and segmentation problem on kaggle (https://www.kaggle.com/c/airbus-ship-detection#description). In this problem, I build a model that detects all ships in satellite images and generate a mask for each ship. There several deep learning models that works with image detection such as YOLO, R-CNN, Fast R-CNN, Faster R-CNN. For objection segmentation, Unet is a great tools. Recently there is a nice paper on object instance segmentation (https://arxiv.org/abs/1703.06870) called Mask R-CNN.

In this problem, most image (~80%) contains no ships. So my strategy is the following:

1. I build a classifier to detect if a image has any ships.
2. Feed the image that contains image detected by the classifier to Mask R-CNN.

Results

Acknowledgement

This code is implemented on maskrcnn frameworks (https://github.com/matterport/Mask_RCNN). Thanks for their great work!

Prerequisites

Python 3.6

Jupyter Notebook

Meta

Chi Zhang – @LinkedIn – c.zhang@neu.edu

Distributed under the MIT license. See LICENSE for more information.