- https://www.kaggle.com/c/dstl-satellite-imagery-feature-detection
- https://www.kaggle.com/alijs1/squeezed-this-in-successful-kernel-run/code
- General overview: https://blog.deepsense.ai/deep-learning-for-satellite-imagery-via-image-segmentation/
-
Data Importing (Goal: 1x 20-channels image)
- Gray-Scale
- 3-Band
- 16-Band
-
Creatable Torch DataLoader
-
Data Transformations
- Up-scaling
- Random-cropping
- Resizing
- Rotations
-
Mask-to-Polygon Transformation (in data_import.py "mask_to_polygons" function)
-
Data visualization
-
Report
- Describing raw data (James)
- Goal & methods/frameworks used (James)
- Implementation/ steps (Philipp)
- Results & difficulties (Philipp)
- Important links
- UNet Implementation
- Model training
- A Review on Deep Learning Techniques Applied to Semantic Segmentation https://arxiv.org/pdf/1704.06857.pdf
- Long, Jonathan, Evan Shelhamer, and Trevor Darrell. "Fully convolutional networks for semantic segmentation." CVPR. 2015. https://arxiv.org/pdf/1605.06211.pdf
- He, Kaiming, et al. "Deep residual learning for image recognition." arXiv:1512.03385, 2015.
- Lee, Chen-Yu, et al. "Deeply-Supervised Nets." AISTATS, 2015.
- Understanding Convolution for Semantic Segmentation, Wang et al
- PSP Net https://arxiv.org/abs/1612.01105
- https://github.com/kjw0612/awesome-deep-vision#semantic-segmentation
- DeepLab: Semantic Image Segmentation with Deep Convolutional Nets, Atrous Convolution and Fully Connected CRFs https://arxiv.org/pdf/1606.00915.pdf