Research Article
[Retracted] Rethinking Separable Convolutional Encoders for End-to-End Semantic Image Segmentation
Table 3
Comparison of experimental results (CamVid validation set).
| Method | Mean IoU | Mean ACC | Overall ACC |
| FCN-8S | 0.4301 | 0.6307 | 0.8347 | FCN-32S | 0.4127 | 0.6425 | 0.8438 | SegNet | 0.3721 | 0.6531 | 0.8271 | SegNet + confrontational learning | 0.3927 | 0.6597 | 0.8364 | SegProNet | 0.5103 | 0.6741 | 0.8521 | SegProNet + confrontational learning | 0.5231 | 0.6872 | 0.8675 | Improvement of SegProNet | 0.6142 | 0.6927 | 0.8843 | Improvement of SegProNet + confrontational learning | 0.6247 | 0.7031 | 0.9021 |
|
|