This is the official implementation of Fine-grained Background Representation for Weakly Supervised Semantic Segmentation that was published in IEEE TCSVT 2024.
Requirements:
Steps of experiments:
- You can directly run the code by executing the script script_contrast.sh on the script folder.
- The steps consist of
- Train the seeds
- Post-processing with CRF
- Evaluation
#Results and trained models
on Pascal Voc 2012
| Method | train set | val set | test set |
|---|---|---|---|
| AMN | 72.2 | 70.7 | 70.6 |
| +ours | 73.1 | 71.8 | 73.2 |
| PPC | 73.6 | 72.6 | 73.6 |
| +ours | 75.9 | 74.2 | 74.9 |
on MS COCO 2014
| Method | train set | val set |
|---|---|---|
| AMN | - | 44.7 |
| +ours | 46.7 | 45.6 |
License:
This project is released under the MIT License.