qubvel/segmentation_models
A Keras/TensorFlow library for image segmentation offering pretrained Unet, FPN, Linknet, and PSPNet architectures with 25 backbone options.
Not currently ranked — collecting fresh signals.
star history
This library provides high-level APIs for building neural network models for binary and multi-class image segmentation. It implements four major segmentation architectures with 25 available pretrained backbones including ResNet, EfficientNet, DenseNet, and MobileNet. The package includes segmentation-specific losses (Jaccard, Dice, Focal) and metrics (IoU, F-score) for training and evaluation.
Frequently asked
- What is qubvel/segmentation_models?
- A Keras/TensorFlow library for image segmentation offering pretrained Unet, FPN, Linknet, and PSPNet architectures with 25 backbone options.
- Is segmentation_models open source?
- Yes — qubvel/segmentation_models is open source, released under the MIT license.
- What language is segmentation_models written in?
- qubvel/segmentation_models is primarily written in Python.
- How popular is segmentation_models?
- qubvel/segmentation_models has 4.9k stars on GitHub.
- Where can I find segmentation_models?
- qubvel/segmentation_models is on GitHub at https://github.com/qubvel/segmentation_models.