Skip to content

CNN-based algorithm for the detection and classification of icebergs and ships in Sentinel-1 SAR (radar satellite) imagery .

Notifications You must be signed in to change notification settings

eikeschuett/IcebergShipDetection

Repository files navigation

IcebergShipDetection

  • Topic: Iceberg and ship detection in satellite imagery
  • Goals: The project goal is to build an algorithm for the detection of ships and icebergs in Sentinel-1 SAR imagery. Desired output is a map, which shows the locations of icebergs, ships and unidentified objects.
  • Details: The dataset used for training is obtained from a Kaggle challenge, Statoil/C-CORE Iceberg Classifier. Each image has 75x75 pixels with two bands from HH and HV polarisations and contains a ship or an iceberg. This dataset will be used to train a CNN. After training the classification model, we will use Sentinel-1 SAR images to show the "real world application" of our model. The satellite images will be pre-processed with the Sentinel Application Platform (SNAP) Python API. We will then identify bright objects within each satellite image. A 75x75 subset of the radar image will be made for each object and fed into our classification model. Finally, the results will be plotted on a map.

Testscenes and AIS Ship Positions

  • Disko Bay
    • S1A_IW_GRDH_1SDH_20210115T100027_20210115T100052_036147_043CF4_049C

      • Other Type/Auxillary N69°08.750' W53°39.933'
      • Fishing Vessel N68°45.080' W51°20.846'
    • S1B_IW_GRDH_1SDH_20210114T100803_20210114T100828_025149_02FE89_234D

      • Other Type/Auxillary N69°05.990' W53°18.654'
      • Fishing Vessel N68°51.626' W52°47.673'
      • Other Type/Auxillary N69°23.972' W51°36.317'
      • Fishing Vessel N68°43.877' W51°30.219'
      • Fishing Vessel N68°43.825' W51°21.114'
      • Cargo Ship N76°28.180' W54°08.052'
  • Svalbard
    • S1B_IW_GRDH_1SDH_20210108T154500_20210108T154525_025065_02FBC6_38D2 (contains only a tanker and no icebergs)
      • Tanker N78°12.417' E14°32.650'

Links

Preprocessing with Snappy

Object Detection and Models

  • Presentation of a detailed Jupyter Notebook with code and comment
    • including the definition of the environment
    • including required sections (Introduction, Data and Methods, Results, Baseline)
  • A small video, accompanying, for example, a screen recording of the notebook with an explanation of the challenge of the project, the used approach, and the results.
  • A statement that the code is released as open source software. The data you use in your project can remain private if you wish.
  • Time: 8 -- 10 minutes

About

CNN-based algorithm for the detection and classification of icebergs and ships in Sentinel-1 SAR (radar satellite) imagery .

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Contributors 3

  •  
  •  
  •