Nearshore Ship Detection on High-Resolution Remote Sensing Image via Scene-Mask R-CNN

Tags: , , ,

Aim:

To automate the detection of presence of ships and to classify the types of ships available in the given Image.

Existing System:

Computer-aided ship detection methods greatly free up human resources and typically include two steps: extracting image features, and then using classifiers for classification and localization. These methods can produce stable results under calm sea conditions. However, when disturbances such as waves, clouds, rain, fog, and reflections happen, the extracted low-level features are not robust. Besides, manual selection of features is time-consuming and strongly dependent on the expertise and characteristics of the data itself.

Proposed System:

A system is proposed to automate the detection of presence of ships in the given image using Machine Learning and Deep Learning Algorithms. We are proposing along with ship detection, a ship classification based on the type and category of the ships. The proposed system will not only detect a ship but also categorize as war ship, container ship etc.