Takikawa T, Acuna D, Jampani V, et al. Gated-scnn: Gated shape cnns for semantic segmentation[C]//Proceedings of the IEEE International Conference on Computer Vision. 2019: 5229-5238.
Model | Backbone | Resolution | Training Iters | mIoU | mIoU (flip) | mIoU (ms+flip) | Links |
---|---|---|---|---|---|---|---|
GSCNN | ResNet50_OS8 | 1024x512 | 80000 | 80.67% | 80.88% | 80.88% | model | log | vdl |