Semantic Segmentation is the computer vision task of assigning a class label to every pixel in an image, partitioning the image into semantically meaningful regions without distinguishing between individual object instances. Architectures such as FCN, U-Net, and DeepLab produce dense pixel-wise predictions enabling scene understanding in autonomous driving, medical imaging, and satellite analysis.
Semantic Classification
Content
Core Characteristics
- Pixel-Level Classification: Dense prediction for every pixel
- Semantic Understanding: Class labels without instance differentiation
- Fully Convolutional: End-to-end architectures without fully connected layers
- Multi-Scale Processing: Encoding-decoding with skip connections
- Contextual Aggregation: Atrous convolution, pyramid pooling
Key Literature
-
- Long, J., Shelhamer, E., & Darrell, T. (2015). “Fully convolutional networks for semantic segmentation.” CVPR, 3431-3440.
-
- Ronneberger, O., Fischer, P., & Brox, T. (2015). “U-Net: Convolutional networks for biomedical image segmentation.” MICCAI, 234-241.
-
- Chen, L. C., et al. (2018). “Encoder-decoder with atrous separable convolution for semantic image segmentation.” ECCV, 801-818.