ERFnet is a semantic segmentation network based on residual connections and separable convolutions designed to improve the efficiency of the model in processing frames without reducing accuracy
| Model | Download | Class IoU (%) | Category IoU (%) |
|---|---|---|---|
| ERFNet | model weight | 69.7 | 87.3 |
cityscapes dataset: website address git address
- ERFNet: Efficient Residual Factorized ConvNet for Real-time Semantic Segmentation
- Yuelong-Yu/ERFNet-Caffe
MIT