PASCAL VOC Challenge performance evaluation and download server |
|
| Home | Leaderboard |
Segmentation Results: VOC2012 BETA
Competition "comp5" (train on VOC2012 data)
This leaderboard shows only those submissions that have been marked as public, and so the displayed rankings should not be considered as definitive.- The highest scoring entry in each column is shown in bold.
- Clicking on the blue arrow symbol (
) at the top of a column will order
the submissions from high to low wrt performance on that column.
Average Precision (AP %)
| mean | aero plane  | bicycle | bird | boat | bottle | bus | car | cat | chair | cow | dining table | dog | horse | motor bike | person | potted plant | sheep | sofa | train | tv/ monitor | submission date | ||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| XC-FLATTENET [?] | 84.3 | 94.0 | 73.2 | 91.5 | 74.4 | 83.2 | 95.5 | 90.5 | 96.7 | 38.1 | 94.5 | 76.2 | 92.9 | 95.6 | 88.9 | 90.4 | 76.0 | 93.8 | 63.6 | 86.8 | 78.6 | 12-Jan-2020 | |
| TUSS-20k [?] | 84.2 | 89.9 | 50.2 | 92.6 | 84.3 | 81.9 | 93.2 | 93.2 | 96.9 | 54.7 | 93.1 | 69.0 | 95.4 | 93.9 | 93.2 | 91.5 | 74.1 | 96.5 | 72.5 | 92.1 | 65.0 | 15-May-2024 | |
| FDNet_16s [?] | 84.0 | 95.4 | 77.9 | 95.9 | 69.1 | 80.6 | 96.4 | 92.6 | 95.5 | 40.5 | 92.6 | 70.6 | 93.8 | 93.1 | 90.4 | 89.9 | 71.2 | 92.7 | 63.1 | 88.5 | 77.7 | 22-Mar-2018 | |
| TUSS-fs [?] | 83.9 | 94.3 | 44.7 | 97.2 | 84.7 | 84.4 | 92.8 | 91.2 | 94.0 | 46.5 | 95.5 | 68.0 | 89.0 | 94.1 | 91.2 | 91.9 | 79.4 | 92.9 | 72.0 | 85.2 | 78.1 | 15-May-2024 | |
| FLATTENET-001 [?] | 83.1 | 95.4 | 69.8 | 86.4 | 73.3 | 78.7 | 94.6 | 92.1 | 95.5 | 41.0 | 92.4 | 75.8 | 93.1 | 95.2 | 89.8 | 90.0 | 65.8 | 94.0 | 60.0 | 86.6 | 79.6 | 29-Dec-2019 | |
| AAA_HEM [?] | 82.9 | 94.5 | 73.6 | 92.4 | 66.5 | 81.7 | 96.3 | 90.0 | 95.3 | 36.2 | 92.1 | 68.1 | 92.4 | 93.2 | 90.5 | 88.9 | 74.1 | 93.9 | 58.5 | 89.7 | 76.6 | 11-Apr-2020 | |
| FS+WSSS [?] | 81.6 | 96.3 | 48.8 | 92.8 | 83.9 | 82.1 | 93.1 | 94.4 | 94.5 | 48.6 | 96.9 | 45.9 | 94.1 | 96.1 | 93.7 | 88.7 | 63.5 | 93.9 | 63.9 | 92.8 | 54.8 | 26-Jun-2023 | |
| TUSS-ws [?] | 81.1 | 85.7 | 45.8 | 97.1 | 75.8 | 84.1 | 92.2 | 92.8 | 93.9 | 43.1 | 95.0 | 49.9 | 90.0 | 95.0 | 90.8 | 91.6 | 77.9 | 92.2 | 60.3 | 87.1 | 68.1 | 15-May-2024 | |
| aux_0.4_finetuning_train_checkpoints [?] | 81.1 | 94.3 | 43.6 | 95.3 | 73.7 | 83.7 | 95.3 | 90.7 | 94.7 | 39.6 | 93.6 | 54.2 | 90.2 | 93.6 | 90.2 | 87.4 | 75.9 | 92.6 | 56.6 | 88.0 | 75.5 | 16-Jun-2020 | |
| WASPNet+CRF [?] | 79.6 | 90.7 | 60.9 | 85.3 | 68.9 | 80.7 | 93.8 | 84.7 | 94.5 | 38.5 | 86.0 | 69.7 | 90.7 | 86.9 | 85.4 | 86.8 | 67.6 | 88.8 | 57.4 | 85.4 | 74.2 | 19-Nov-2019 | |
| WASPNet [?] | 79.4 | 89.5 | 63.8 | 87.6 | 68.8 | 79.1 | 93.5 | 84.7 | 93.7 | 37.6 | 84.8 | 69.4 | 90.8 | 87.9 | 85.7 | 86.5 | 67.4 | 87.0 | 57.2 | 85.2 | 72.7 | 25-Jul-2019 | |
| acm [?] | 79.0 | 90.9 | 41.6 | 91.1 | 65.2 | 80.7 | 93.6 | 90.0 | 92.7 | 39.4 | 88.6 | 60.8 | 88.6 | 89.7 | 88.8 | 85.6 | 72.1 | 89.4 | 58.2 | 84.9 | 72.2 | 13-Jun-2020 | |
| Unet_Attention [?] | 78.0 | 93.0 | 67.9 | 83.6 | 65.9 | 78.6 | 92.4 | 87.3 | 90.0 | 30.5 | 82.5 | 65.8 | 86.2 | 83.9 | 86.7 | 85.3 | 66.2 | 86.8 | 55.7 | 82.1 | 73.1 | 17-Jan-2022 | |
| GDM [?] | 74.5 | 85.8 | 40.2 | 81.6 | 61.8 | 67.3 | 90.8 | 84.2 | 91.8 | 37.3 | 80.2 | 62.7 | 85.2 | 86.4 | 85.9 | 81.7 | 56.5 | 86.6 | 56.7 | 79.7 | 69.4 | 04-Mar-2020 | |
| refinenet_HPM [?] | 74.2 | 87.9 | 62.2 | 76.0 | 55.3 | 76.0 | 86.7 | 82.6 | 85.4 | 28.9 | 79.6 | 64.2 | 81.2 | 79.4 | 85.1 | 85.0 | 65.6 | 83.2 | 51.3 | 79.2 | 68.6 | 01-Mar-2019 | |
| CDL_new [?] | 73.1 | 89.4 | 38.9 | 87.4 | 63.6 | 75.1 | 90.7 | 83.6 | 89.2 | 35.8 | 83.4 | 42.7 | 85.0 | 87.6 | 82.2 | 81.5 | 57.4 | 85.9 | 48.2 | 82.1 | 53.2 | 29-Jun-2022 | |
| DCONV_SSD_FCN [?] | 72.9 | 88.6 | 38.1 | 85.2 | 57.8 | 71.4 | 90.8 | 84.2 | 86.0 | 32.1 | 83.4 | 53.7 | 80.4 | 80.8 | 81.6 | 81.2 | 61.4 | 84.1 | 51.9 | 77.5 | 67.0 | 17-Mar-2018 | |
| Puzzle-CAM [?] | 72.3 | 87.2 | 37.4 | 86.8 | 61.5 | 71.3 | 92.2 | 86.3 | 91.8 | 28.6 | 85.1 | 64.2 | 91.9 | 82.1 | 82.6 | 70.7 | 69.4 | 87.7 | 45.5 | 67.0 | 37.8 | 02-Feb-2021 | |
| CDL_VWL [?] | 71.7 | 87.6 | 33.7 | 89.8 | 60.1 | 68.3 | 91.4 | 83.1 | 89.9 | 33.6 | 78.3 | 62.0 | 84.4 | 82.5 | 82.5 | 78.7 | 42.7 | 81.4 | 60.0 | 58.6 | 65.3 | 12-Jul-2022 | |
| Deeplab-clims-r38 [?] | 71.2 | 82.1 | 34.0 | 85.7 | 58.2 | 70.5 | 87.0 | 82.3 | 86.3 | 30.8 | 80.9 | 59.4 | 83.1 | 80.7 | 80.6 | 78.4 | 54.5 | 85.3 | 59.5 | 69.8 | 53.6 | 25-Oct-2022 | |
| CGPT [?] | 70.8 | 77.0 | 32.1 | 84.1 | 56.4 | 66.5 | 87.6 | 83.1 | 87.8 | 29.8 | 81.5 | 59.6 | 84.9 | 84.0 | 80.2 | 77.3 | 50.4 | 81.4 | 61.2 | 72.8 | 57.6 | 22-May-2023 | |
| CGPT_IMN [?] | 70.4 | 79.0 | 32.1 | 85.5 | 58.9 | 65.5 | 86.4 | 81.2 | 87.1 | 27.1 | 82.4 | 57.0 | 85.0 | 82.6 | 79.7 | 76.1 | 49.3 | 83.5 | 58.4 | 73.2 | 57.5 | 22-May-2023 | |
| CDL_new_rib [?] | 69.9 | 85.0 | 34.8 | 87.9 | 59.9 | 75.5 | 90.6 | 81.6 | 89.1 | 33.2 | 81.1 | 39.6 | 83.9 | 82.9 | 81.7 | 74.9 | 37.8 | 82.7 | 57.5 | 58.3 | 58.5 | 29-Jun-2022 | |
| AD_CLIMS [?] | 69.8 | 81.0 | 33.0 | 88.1 | 60.6 | 69.0 | 87.9 | 81.7 | 88.9 | 27.5 | 82.5 | 60.2 | 85.8 | 83.5 | 78.9 | 34.2 | 52.1 | 80.5 | 58.5 | 79.1 | 61.9 | 03-Mar-2023 | |
| deeplabv3_plus_reproduction [?] | 69.5 | 81.7 | 38.1 | 83.1 | 60.1 | 62.7 | 89.8 | 81.4 | 87.6 | 30.0 | 72.8 | 60.9 | 78.1 | 77.7 | 78.6 | 78.1 | 44.9 | 76.7 | 50.8 | 75.4 | 58.6 | 11-May-2019 | |
| attention RRM [?] | 69.2 | 85.2 | 31.2 | 86.6 | 49.3 | 71.9 | 85.5 | 76.1 | 88.8 | 32.6 | 76.8 | 64.0 | 84.5 | 82.6 | 76.9 | 74.3 | 47.5 | 82.7 | 57.8 | 42.4 | 65.6 | 12-Mar-2021 | |
| Progressive Framework [?] | 69.2 | 83.4 | 35.7 | 82.5 | 46.0 | 66.0 | 88.4 | 79.1 | 90.9 | 28.6 | 82.4 | 44.3 | 85.9 | 84.7 | 78.5 | 79.1 | 53.4 | 83.3 | 44.5 | 61.3 | 63.4 | 06-Jul-2021 | |
| DeeplabV3+_Exploring_Missing_Parts [?] | 68.8 | 88.7 | 37.1 | 87.6 | 57.9 | 64.3 | 85.9 | 73.2 | 85.4 | 24.7 | 84.2 | 29.4 | 81.8 | 90.2 | 81.8 | 77.3 | 55.5 | 88.0 | 34.9 | 74.6 | 51.3 | 15-Jul-2021 | |
| TCnet [?] | 68.4 | 72.6 | 32.6 | 74.2 | 59.5 | 68.9 | 86.7 | 77.1 | 78.6 | 34.4 | 68.6 | 63.0 | 74.4 | 75.8 | 76.3 | 77.0 | 54.9 | 76.6 | 55.5 | 76.9 | 61.7 | 02-May-2018 | |
| UMICH_EG-ConvCRF_Iter_Res101 [?] | 67.9 | 76.4 | 32.1 | 82.0 | 52.9 | 69.5 | 89.5 | 81.1 | 87.5 | 29.6 | 74.7 | 54.2 | 83.4 | 79.2 | 78.7 | 76.1 | 48.1 | 81.0 | 59.4 | 51.3 | 49.3 | 13-Dec-2019 | |
| UMICH_TCS_101 [?] | 66.7 | 74.6 | 31.5 | 81.5 | 50.3 | 67.2 | 88.4 | 80.6 | 82.3 | 29.9 | 72.3 | 53.4 | 76.1 | 76.0 | 78.4 | 73.6 | 47.5 | 79.2 | 52.0 | 57.9 | 56.6 | 01-Dec-2019 | |
| UMICH_EG-ConvCRF_Iter_Res50 [?] | 66.4 | 75.8 | 32.2 | 85.5 | 50.9 | 69.3 | 86.4 | 79.6 | 85.1 | 29.1 | 73.7 | 55.7 | 79.6 | 74.7 | 76.3 | 75.8 | 44.8 | 79.6 | 51.0 | 50.5 | 48.5 | 10-Dec-2019 | |
| UMICH_TCS [?] | 65.5 | 73.6 | 32.4 | 81.6 | 50.4 | 68.5 | 86.2 | 79.4 | 81.8 | 28.2 | 75.5 | 55.6 | 79.0 | 75.5 | 77.7 | 74.3 | 48.2 | 79.0 | 52.3 | 44.4 | 42.7 | 28-Nov-2019 | |
| bothweight th0.4 [?] | 65.3 | 79.1 | 33.4 | 88.2 | 20.1 | 65.3 | 88.0 | 76.2 | 90.0 | 24.7 | 80.7 | 43.7 | 85.1 | 85.8 | 82.3 | 69.8 | 47.7 | 84.9 | 43.8 | 41.4 | 54.5 | 15-Apr-2019 | |
| weight+RS [?] | 64.5 | 85.0 | 31.9 | 85.4 | 19.1 | 65.3 | 88.6 | 72.9 | 88.5 | 24.8 | 75.6 | 50.1 | 83.3 | 82.0 | 81.6 | 66.8 | 56.6 | 80.3 | 45.8 | 44.6 | 39.7 | 24-Mar-2019 | |
| AttnBN [?] | 63.0 | 75.7 | 32.9 | 73.5 | 49.9 | 60.4 | 78.1 | 76.5 | 77.4 | 19.9 | 72.0 | 27.4 | 73.8 | 72.7 | 77.2 | 72.3 | 51.2 | 77.3 | 37.9 | 73.5 | 53.6 | 14-Aug-2019 | |
| DSRG_ATTNBN [?] | 63.0 | 76.9 | 32.3 | 72.9 | 49.0 | 59.2 | 77.7 | 75.4 | 76.7 | 19.5 | 71.5 | 27.9 | 74.3 | 73.6 | 77.0 | 72.8 | 52.7 | 76.4 | 40.3 | 73.5 | 52.7 | 28-Feb-2020 | |
| Progressive Framework [?] | 60.0 | 72.2 | 32.4 | 71.9 | 42.4 | 63.2 | 70.2 | 70.3 | 77.8 | 23.4 | 60.5 | 33.2 | 72.4 | 71.0 | 75.4 | 69.8 | 39.9 | 70.2 | 37.7 | 64.0 | 53.2 | 06-Jul-2021 | |
| Extended [?] | 59.3 | 77.9 | 28.9 | 75.1 | 42.6 | 55.2 | 70.4 | 58.9 | 53.0 | 24.3 | 66.7 | 51.9 | 73.1 | 71.3 | 72.5 | 63.9 | 45.2 | 59.2 | 43.9 | 65.2 | 58.6 | 29-Aug-2018 | |
| weakly_seg_validation_test [?] | 57.7 | 67.6 | 31.1 | 66.4 | 41.9 | 60.1 | 70.6 | 65.4 | 71.8 | 25.3 | 63.6 | 24.7 | 72.2 | 68.7 | 68.3 | 68.8 | 41.6 | 67.5 | 33.6 | 65.0 | 49.2 | 08-Sep-2019 | |
| Progressive Framework [?] | 55.0 | 64.9 | 30.8 | 68.4 | 31.6 | 52.6 | 70.5 | 64.8 | 73.4 | 22.6 | 48.6 | 33.4 | 68.6 | 59.6 | 71.1 | 68.9 | 38.5 | 60.7 | 40.8 | 47.7 | 51.7 | 03-Jul-2021 | |
| fcn [?] | 51.0 | 57.0 | 6.2 | 55.0 | 34.9 | 51.0 | 69.3 | 67.8 | 66.7 | 13.7 | 46.5 | 47.0 | 54.9 | 52.4 | 59.8 | 64.8 | 34.4 | 58.9 | 35.6 | 58.7 | 47.5 | 26-Apr-2023 | |
| O2P_SVRSEGM_CPMC_CSI [?] | 47.5 | 64.0 | 32.2 | 45.9 | 34.7 | 46.3 | 59.5 | 61.7 | 49.4 | 14.8 | 47.9 | 31.2 | 42.5 | 51.3 | 58.8 | 54.6 | 34.9 | 54.6 | 34.7 | 50.6 | 42.2 | 15-Nov-2012 | |
| NUS_DET_SPR_GC_SP [?] | 47.3 | 52.9 | 31.0 | 39.8 | 44.5 | 58.9 | 60.8 | 52.5 | 49.0 | 22.6 | 38.1 | 27.5 | 47.4 | 52.4 | 46.8 | 51.9 | 35.7 | 55.3 | 40.8 | 54.2 | 47.8 | 23-Sep-2012 | |
| BONN_O2PCPMC_FGT_SEGM [?] | 47.0 | 65.4 | 29.3 | 51.3 | 33.4 | 44.2 | 59.8 | 60.3 | 52.5 | 13.6 | 53.6 | 32.6 | 40.3 | 57.6 | 57.3 | 49.0 | 33.5 | 53.5 | 29.2 | 47.6 | 37.6 | 23-Sep-2012 | |
| vgg_unet [?] | 45.8 | 56.3 | 44.6 | 54.6 | 37.0 | 44.1 | 55.8 | 55.2 | 55.8 | 13.1 | 30.7 | 32.8 | 45.4 | 39.9 | 56.9 | 65.9 | 27.3 | 51.1 | 23.3 | 44.4 | 38.7 | 13-Aug-2023 | |
| BONNGC_O2P_CPMC_CSI [?] | 45.4 | 59.3 | 27.9 | 43.9 | 39.8 | 41.4 | 52.2 | 61.5 | 56.4 | 13.6 | 44.5 | 26.1 | 42.8 | 51.7 | 57.9 | 51.3 | 29.8 | 45.7 | 28.8 | 49.9 | 43.3 | 23-Sep-2012 | |
| BONN_CMBR_O2P_CPMC_LIN [?] | 44.8 | 60.0 | 27.3 | 46.4 | 40.0 | 41.7 | 57.6 | 59.0 | 50.4 | 10.0 | 41.6 | 22.3 | 43.0 | 51.7 | 56.8 | 50.1 | 33.7 | 43.7 | 29.5 | 47.5 | 44.7 | 23-Sep-2012 | |
| fcn [?] | 38.4 | 51.6 | - | 11.8 | 23.9 | 36.8 | 65.7 | 55.7 | 57.8 | 3.8 | 33.5 | 41.2 | 40.8 | 39.4 | 42.6 | 52.7 | 12.7 | 26.1 | 26.5 | 56.5 | 42.4 | 26-Apr-2023 | |
| comp6_test_cls [?] | 37.7 | 36.6 | 10.8 | 38.9 | 25.9 | 30.8 | 56.0 | 53.8 | 57.8 | 4.9 | 24.6 | 22.1 | 48.1 | 33.1 | 32.6 | 56.1 | 23.5 | 29.7 | 31.8 | 42.7 | 45.6 | 10-May-2018 | |
| OptNBNN-CRF [?] | 11.3 | 10.5 | 2.3 | 3.0 | 3.0 | 1.0 | 30.2 | 14.9 | 15.0 | 0.2 | 6.1 | 2.3 | 5.1 | 12.1 | 15.3 | 23.4 | 0.5 | 8.9 | 3.5 | 10.7 | 5.3 | 23-Sep-2012 | |
| unet_resnet50 [?] | - | 62.9 | 36.3 | 57.0 | 38.9 | 57.0 | 67.6 | 70.9 | 65.9 | - | 24.6 | 39.5 | 61.6 | 39.3 | 59.3 | 65.7 | 26.1 | 49.1 | 29.4 | 56.4 | 51.0 | 13-Aug-2023 |
Abbreviations
| Title | Method | Affiliation | Contributors | Description | Date |
|---|---|---|---|---|---|
| test | AAA_HEM | xiongdeng@stu.xmu.edu.cn | 111 | test | 2020-04-11 03:21:43 |
| Clims with adapter | AD_CLIMS | ETS, Montreal | BM, RH, RB | CLIMS in WSS settings | 2023-03-03 17:57:45 |
| AttnBN | AttnBN | AttnBN | AttnBN | AttnBN | 2019-08-14 23:23:24 |
| O2P Regressor + Composite Statistical Inference | BONNGC_O2P_CPMC_CSI | (1) University of Bonn, (2) Georgia Institute of Technology, (3) University of Coimbra | Joao Carreira (1,3) Fuxin Li (2) Guy Lebanon (2) Cristian Sminchisescu (1) | We utilize a novel probabilistic inference procedure (unpublished yet), Composite Statisitcal Inference (CSI), on semantic segmentation using predictions on overlapping figure-ground hypotheses. Regressor predictions on segment overlaps to the ground truth object are modelled as generated by the true overlap with the ground truth segment plus noise. A model of ground truth overlap is defined by parametrizing on the unknown percentage of each superpixel that belongs to the unknown ground truth. A joint optimization on all the superpixels and all the categories is then performed in order to maximize the likelihood of the SVR predictions. The optimization has a tight convex relaxation so solutions can be expected to be close to the global optimum. A fast and optimal search algorithm is then applied to retrieve each object. CSI takes the intuition from the SVRSEGM inference algorithm that multiple predictions on similar segments can be combined to better consolidate the segment mask. But it fully develops the idea by constructing a probabilistic framework and performing composite MLE jointly on all segments and categories. Therefore it is able to consolidate better object boundaries and handle hard cases when objects interact closely and heavily occlude each other. For each image, we use 150 overlapping figure-ground hypotheses generated by the CPMC algorithm (Carreira and Sminchisescu, PAMI 2012), and linear SVR predictions on them with the novel second order O2P features (Carreira, Caseiro, Batista, Sminchisescu, ECCV2012; see VOC12 entry BONN_CMBR_O2P_CPMC_LIN) as the input to the inference algorithm. | 2012-09-23 23:49:02 |
| Linear SVR with second-order pooling. | BONN_CMBR_O2P_CPMC_LIN | (1) University of Bonn, (2) University of Coimbra | Joao Carreira (2,1) Rui Caseiro (2) Jorge Batista (2) Cristian Sminchisescu (1) | We present a novel effective local feature aggregation method that we use in conjunction with an existing figure-ground segmentation sampling mechanism. This submission is described in detail in [1]. We sample multiple figure-ground segmentation candidates per image using the Constrained Parametric Min-Cuts (CPMC) algorithm. SIFT, masked SIFT and LBP features are extracted on the whole image, then pooled over each object segmentation candidate to generate global region descriptors. We employ a novel second-order pooling procedure, O2P, with two non-linearities: a tangent space mapping and power normalization. The global region descriptors are passed through linear regressors for each category, then labeled segments in each image having scores above some threshold are pasted onto the image in the order of these scores. Learning is performed using an epsilon-insensitive loss function on overlap with ground truth, similar to [2], but within a linear formulation (using LIBLINEAR). comp6: learning uses all images in the segmentation+detection trainval sets, and external ground truth annotations provided by courtesy of the Berkeley vision group. comp5: one model is trained for each category using the available ground truth segmentations from the 2012 trainval set. Then, on each image having no associated ground truth segmentations, the learned models are used together with bounding box constraints, low-level cues and region competition to generate predicted object segmentations inside all bounding boxes. Afterwards, learning proceeds similarly to the fully annotated case. 1. “Semantic Segmentation with Second-Order Pooling”, Carreira, Caseiro, Batista, Sminchisescu. ECCV 2012. 2. "Object Recognition by Ranking Figure-Ground Hypotheses", Li, Carreira, Sminchisescu. CVPR 2010. | 2012-09-23 19:11:47 |
| BONN_O2PCPMC_FGT_SEGM | BONN_O2PCPMC_FGT_SEGM | (1) Universitfy of Bonn, (2) University of Coimbra, (3) Georgia Institute of Technology, (4) Vienna University of Technology | Joao Carreira(1,2), Adrian Ion(4), Fuxin Li(3), Cristian Sminchisescu(1) | We present a joint image segmentation and labeling model which, given a bag of figure-ground segment hypotheses extracted at multiple image locations and scales using CPMC (Carreira and Sminchisescu, PAMI 2012), constructs a joint probability distribution over both the compatible image interpretations (tilings or image segmentations) composed from those segments, and over their labeling into categories. The process of drawing samples from the joint distribution can be interpreted as first sampling tilings, modeled as maximal cliques, from a graph connecting spatially non-overlapping segments in the bag (Ion, Carreira, Sminchisescu, ICCV2011), followed by sampling labels for those segments, conditioned on the choice of a particular tiling. We learn the segmentation and labeling parameters jointly, based on Maximum Likelihood with a novel Incremental Saddle Point estimation procedure (Ion, Carreira, Sminchisescu, NIPS2011). As meta-features we combine outputs from linear SVRs using novel second order O2P features to predict the overlap between segments and ground-truth objects of each class (Carreira, Caseiro, Batista, Sminchisescu, ECCV2012; see VOC12 entry BONNCMBR_O2PCPMC_LINEAR), bounding box object detectors, and kernel SVR outputs trained to predict the overlap between segments and ground-truth objects of each class (Carreira, Li, Sminchisescu, IJCV 2012). comp6: the O2P SVR learning uses all images in the segmentation+detection trainval sets, and external ground truth annotations provided by courtesy of the Berkeley vision group. | 2012-09-23 21:39:35 |
| CDL_VWL | CDL_VWL | XJTLU;UoL | Bingfeng Zhang Jimin Xiao | VML IJCV22 WSSS COCO-PRE | 2022-07-12 02:46:57 |
| CDL_NEW | CDL_new | XJTLU | Bingfeng Zhang Jimin Xiao | wsss EPS | 2022-06-29 03:31:34 |
| CDL_NEW_rib | CDL_new_rib | XJTLU | Bingfeng Zhang Jimin Xiao | WSSS RIB | 2022-06-29 03:47:40 |
| WSS Segmentor | CGPT | ETS, Montreal | RH, RB, BM, JD | Segmentation using WSS | 2023-05-22 07:14:39 |
| WSS Seg | CGPT_IMN | ETS, Montreal | RB, RH, BM, JD | WSS Segmentation | 2023-05-22 13:55:35 |
| dssd style arch | DCONV_SSD_FCN | shanghai university | li junhao(jxlijunhao@163.com) | combine object detection and semantic segmentation in one forward pass | 2018-03-17 02:58:20 |
| DSRG_ATTNBN | DSRG_ATTNBN | DSRG_ATTNBN | DSRG_ATTNBN | DSRG_ATTNBN | 2020-02-28 08:27:04 |
| Deeplab-clims-r38 | Deeplab-clims-r38 | Ecole de technologie superieure | Balamurali Murugesan | Deeplab-clims-r38 | 2022-10-25 21:52:00 |
| Exploring the missing parts for WSSS | DeeplabV3+_Exploring_Missing_Parts | Northestern University | Dali Chen | Exploring the mssing parts in the muti-categories saliency map and improve the accuracy of all of the WSSS methods based on saliency map. | 2021-07-15 06:59:42 |
| Weakly-supervised model | Extended | SYSU | Wenfeng Luo | DCNN trained under image labels | 2018-08-29 12:54:29 |
| FDNet_16s | FDNet_16s | HongKong University of Science and Technology, altizure.com | Mingmin Zhen, Jinglu Wang, Siyu Zhu, Runze Zhang, Shiwei Li, Tian Fang, Long Quan | A fully dense neural network with encoder-decoder structure is proposed that we abbreviate as FDNet. For each stage in the decoder module, feature maps of all the previous blocks are adaptively aggregated to feedforward as input. | 2018-03-22 08:52:44 |
| Fully Convolutional Network | FLATTENET-001 | Sichuan University, China | Xin Cai | In contrast to the commonly-used strategies, such as dilated convolution and encoder-decoder structure, we introduce the Flattening Module to produce high-resolution predictions without either removing any subsampling operations or building a complicated decoder module. https://arxiv.org/abs/1909.09961 | 2019-12-29 07:29:05 |
| Foundation Model Assisted Weakly Supervised Semant | FS+WSSS | Zhejiang University | Xiaobo Yang | Foundation Model Assisted Weakly Supervised Semant | 2023-06-26 15:53:48 |
| Global Distinguishing Module | GDM | Jinan University | Runkai Zheng | Uncertainty weighted loss for extracting globally distinguishable spatial features. | 2020-03-04 10:42:09 |
| DM2: Detection, Mask transfer, MRF pruning | NUS_DET_SPR_GC_SP | National University of Singapore(NUS), Panasonic Singapore Laboratories(PSL) | (NUS) Wei XIA, Csaba DOMOKOS, Jian DONG, Shuicheng YAN, Loong Fah CHEONG, (PSL) Zhongyang HUANG, Shengmei SHEN | We propose a three-step coarse-to-fine framework for general object segmentation. Given a test image, the object bounding boxes are first predicted by object detectors, and then the coarse masks within the corresponding bounding boxes are transferred from the training data based on the optimization framework of coupled global and local sparse representations in [1]. Then based on the coarse masks as well as the original detection information (bounding boxes and confidence maps), we built a super-pixel based MRF model for each bounding box, and then perform foreground-background inference. Both L-a-b color histogram and detection confidence map are used for characterizing the unary terms, while the PB edge contrast is used as smoothness term. Finally, the segmentation results are further refined by post-processing of multi-scale super-pixel segmentation. [1]Wei Xia, Zheng Song, Jiashi Feng, Loong Fah Cheong and Shuicheng Yan. Segmentation over Detection by Coupled Global and Local Sparse Representations, ECCV 2012. | 2012-09-23 20:01:56 |
| O2P+SVRSEGM Regressor + Composite Statistical Inference | O2P_SVRSEGM_CPMC_CSI | (1) Georgia Institute of Technology (2) University of California - Berkeley (3) Amazon Inc. (4) Lund University | Fuxin Li | We utilize a novel probabilistic inference procedure, Composite Statisitcal Inference (CSI) [1], on semantic segmentation using predictions on overlapping figure-ground hypotheses. Regressor predictions on segment overlaps to the ground truth object are modelled as generated by the true overlap with the ground truth segment plus noise, parametrized on the unknown percentage of each superpixel that belongs to the unknown ground truth. A joint optimization on all the superpixels and all the categories is then performed in order to maximize the likelihood of the SVR predictions. The optimization has a tight convex relaxation so solutions can be expected to be close to the global optimum. A fast and optimal search algorithm is then applied to retrieve each object. CSI takes the intuition from the SVRSEGM inference algorithm that multiple predictions on similar segments can be combined to better consolidate the segment mask. But it fully develops the idea by constructing a probabilistic framework and performing maximum composite likelihood jointly on all segments and categories. Therefore it is able to consolidate better object boundaries and handle hard cases when objects interact closely and heavily occlude each other. For each image, we use 150 overlapping figure-ground hypotheses generated by the CPMC algorithm (Carreira and Sminchisescu, PAMI 2012), SVRSEGM results, and linear SVR predictions on them with the novel second order O2P features (Carreira, Caseiro, Batista, Sminchisescu, ECCV2012; see VOC12 entry BONN_CMBR_O2P_CPMC_LIN) as the input to the inference algorithm. [1] Fuxin Li, Joao Carreira, Guy Lebanon, Cristian Sminchisescu. Composite Statistical Inference for Semantic Segmentation. CVPR 2013. | 2012-11-15 22:50:41 |
| CRF with NBNN features and simple smoothing | OptNBNN-CRF | University of Amsterdam (UvA) | Carsten van Weelden, Maarten van der Velden, Jan van Gemert | Naive Bayes nearest neighbor (NBNN) [Boiman et al, CVPR 2008] performs well in image classification because it avoids quantization of image features and estimates image-to-class distance. In the context of my MSc thesis we applied the NBNN method to segmentation by estimating image-to-class distances for superpixels, which we use as unary potentials in a simple conditional random field (CRF). To get the NBNN estimates we extract dense SIFT features from the training set and store these in a FLANN index [Muja and Lowe, VISSAPP'09] for efficient nearest neighbor search. To deal with the unbalanced class frequency we learn a linear correction for each class as in [Behmo et al, ECCV 2010]. We segment each test image into 500 SLIC superpixels [Achanta et al, TPAMI 2012] and take each superpixel as a vertex in the CRF. We use the corrected NBNN estimates as unary potentials and Potts potential as pairwise potentials and infer the MAP labeling using alpha-expansion [Boykov et al, TPAMI 2001]. We tune the weighting between the unary and pairwise potential by exhaustive search. | 2012-09-23 12:48:10 |
| Progressive Framework using Weak Autoencoder (SEC) | Progressive Framework | Lakehead University [1,2], Vector Institute [2], The University of British Columbia (Okanagan) [3], Nanyang Technological University [4] | Terence Chow [1] (ychow@lakeheadu.ca) Yimin Yang [2] (yyang48@lakeheadu.ca) Shan Du [3] (shan.du@ubc.ca) Zhiping Linc [4] (EZPLin@ntu.edu.sg) | http://www.yiminyang.com/weakly_supervised.html | 2021-07-03 15:40:47 |
| Progressive Framework using Weak Autoencoder(DSRG) | Progressive Framework | Lakehead University [1,2], Vector Institute [2], The University of British Columbia (Okanagan) [3], Nanyang Technological University [4] | Terence Chow [1] (ychow@lakeheadu.ca) Yimin Yang [2] (yyang48@lakeheadu.ca) Shan Du [3] (shan.du@ubc.ca) Zhiping Linc [4] (EZPLin@ntu.edu.sg) | http://www.yiminyang.com/weakly_supervised.html | 2021-07-06 02:51:33 |
| Progressive Framework using Weak Autoencoder(CIAN) | Progressive Framework | Lakehead University [1,2], Vector Institute [2], The University of British Columbia (Okanagan) [3], Nanyang Technological University [4] | Terence Chow [1] (ychow@lakeheadu.ca) Yimin Yang [2] (yyang48@lakeheadu.ca) Shan Du [3] (shan.du@ubc.ca) Zhiping Linc [4] (EZPLin@ntu.edu.sg) | http://www.yiminyang.com/weakly_supervised.html | 2021-07-06 02:52:28 |
| Puzzle-CAM with ResNeSt-269 | Puzzle-CAM | GYNetworks | Sanghyun Jo, In-Jae Yu | Weakly-supervised semantic segmentation (WSSS) is introduced to narrow the gap for semantic segmentation performance from pixel-level supervision to image-level supervision. Most advanced approaches are based on class activation maps (CAMs) to generate pseudo-labels to train the segmentation network. The main limitation of WSSS is that the process of generating pseudo-labels from CAMs that use an image classifier is mainly focused on the most discriminative parts of the objects. To address this issue, we propose Puzzle-CAM, a process that minimizes differences between the features from separate patches and the whole image. Our method consists of a puzzle module and two regularization terms to discover the most integrated region in an object. Puzzle-CAM can activate the overall region of an object using image-level supervision without requiring extra parameters. % In experiments, Puzzle-CAM outperformed previous state-of-the-art methods using the same labels for supervision on the PASCAL VOC 2012 test dataset. In experiments, Puzzle-CAM outperformed previous state-of-the-art methods using the same labels for supervision on the PASCAL VOC 2012 dataset. | 2021-02-02 05:25:31 |
| TCnet | TCnet | Tsinghua University | Liu Yulin | TCnet | 2018-05-02 08:02:45 |
| TUSS-20k | TUSS-20k | Zhejiang University | Xiaobo Yang | TUSS-20k | 2024-05-15 16:07:14 |
| TUSS-fs | TUSS-fs | Zhejiang University | TUSS-fs | TUSS-fs | 2024-05-15 16:12:32 |
| TUSS-ws | TUSS-ws | Zhejiang University | TUSS-ws | TUSS-ws | 2024-05-15 16:11:57 |
| Iterative method with Entropy-gated ConvCRF | UMICH_EG-ConvCRF_Iter_Res101 | University of Michigan Deep Learning Research Group | Chuan Cen, supervisor: Prof. Honglak Lee | Train segmentation network and relation models iteratively. Infer pseudo-labels for segmentation network with the novel Entropy-gated ConvCRF, which is proved to be superior to random walk under the same conditions. Seg net: Deeplabv2 Seg net backbone: Res101 Relation model backbone: Res101 | 2019-12-13 00:37:25 |
| Iterative method with Entropy-gated ConvCRF | UMICH_EG-ConvCRF_Iter_Res50 | University of Michigan Deep Learning Research Group | Chuan Cen, supervisor: Prof. Honglak Lee | Train segmentation network and relation models iteratively. Infer pseudo-labels for segmentation network with the novel Entropy-gated ConvCRF, which is proved to be superior to random walk under the same conditions. Seg net: Deeplabv2 Seg net backbone: Res50 Relation model backbone: Res50 | 2019-12-10 22:02:49 |
| Transductive semi-sup, co-train, self-train | UMICH_TCS | University of Michigan Deep Learning Research Group | Chuan Cen | It's a method for solving weakly supervised semantic segmentation problem with image-level label only. The problem is viewed as a semi-supervised learning task, then apply graph semi-supervised learning method, co-training and self-training methods together achieving the SOTA performance. | 2019-11-28 19:09:22 |
| Transductive semi-sup, co-train, self-train | UMICH_TCS_101 | University of Michigan Deep Learning Research Group | Chuan Cen | It's a method for solving weakly supervised semantic segmentation problem with image-level label only. The problem is viewed as a semi-supervised learning task, then apply graph semi-supervised learning method, co-training and self-training methods together achieving the SOTA performance. | 2019-12-01 02:40:47 |
| PRETRAINED resnetv2_50x1_bit_distilled (384x384 si | Unet_Attention | HSE | Ivan Vassilenko | PRETRAINED resnetv2_50x1_bit_distilled (384x384 size) | 2022-01-17 07:57:57 |
| WASP for Effective Semantic Segmentation | WASPNet | Rochester Institute of Technology | Bruno Artacho and Andreas Savakis, Rochester Institute of Technology | We propose an efficient architecture for semantic segmentation based on an improvement of Atrous Spatial Pyramid Pooling that achieves a considerable accuracy increase while decreasing the number of parameters and amount of memory necessary. Current semantic segmentation methods rely either on deconvolutional stages that inherently require a large number of parameters, or cascade methods that abdicate larger fields-of-views obtained in the parallelization. The proposed Waterfall architecture leverages the progressive information abstraction from cascade architecture while obtaining multi-scale fields-of-view from spatial pyramid configurations. We demonstrate that the Waterfall approach is a robust and efficient architecture for semantic segmentation using ResNet type networks and obtaining state-of-the-art results with over 20% reduction in the number of parameters and improved performance. | 2019-07-25 20:28:04 |
| Waterfall Atrous Spatial Pooling Arch. for Sem Seg | WASPNet+CRF | Rochester Institute of Technology | Rochester Institute of Technology Bruno Artacho Andreas Savakis | We propose a new efficient architecture for semantic segmentation based on a "Waterfall" Atrous Spatial Pooling architecture that achieves a considerable accuracy increase while decreasing the number of network parameters and memory footprint. The proposed Waterfall architecture leverages the efficiency of progressive filtering in the cascade architecture, while maintaining multi-scale fields-of-view comparable to spatial pyramid configurations. Additionally, our method does not rely on a post-processing stage with Conditional Random Fields, which further reduces complexity and required training time. We demonstrate that the Waterfall approach with a ResNet backbone is a robust and efficient architecture for semantic segmentation obtaining state-of-the-art results with significant reduction in the number of parameters for the Pascal VOC dataset and the Cityscapes dataset. | 2019-11-19 15:19:18 |
| FLATTENET | XC-FLATTENET | Sichuan University, China | Xin Cai | It is well-known that the reduced feature resolution due to repeated subsampling operations poses a serious challenge to Fully Convolutional Network (FCN) based models. In contrast to the commonly-used strategies, such as dilated convolution and encoder-decoder structure, we introduce a novel Flattening Module to produce high-resolution predictions without either removing any subsampling operations or building a complicated decoder module. https://ieeexplore.ieee.org/document/8932465/metrics#metrics | 2020-01-12 02:43:17 |
| Improved deeplabv3 for semantic segmentation | acm | Tianjin University of Technology and Education | *** | **** | 2020-06-13 17:49:57 |
| attention RRM | attention RRM | XJTLU | Bingfeng Zhang Jimin Xiao | RFAM RRM weakly supervised semantic segmentation image level | 2021-03-12 11:49:44 |
| ConvNet for voc 2012 | aux_0.4_finetuning_train_checkpoints | HoHai University | wuY | sda | 2020-06-16 01:45:47 |
| bothweight th0.4 | bothweight th0.4 | Northwestern Politechnical University | Peng Wang, Chunhua Shen | bothweight th0.4 27082 | 2019-04-15 10:46:51 |
| comp6_test_cls | comp6_test_cls | comp6_test_cls | comp6_test_cls | comp6_test_cls | 2018-05-10 15:54:47 |
| pretrained resnet_101 and ASPP module | deeplabv3_plus_reproduction | Institute of Computing Technology | Zhu Lifa | Reproduction of deeplabv3plus with Tensorflow. | 2019-05-11 09:45:53 |
| fcn | fcn | golangboy | golangboy | fcn | 2023-04-26 02:00:55 |
| fcn | fcn | golangboy | golangboy | fcn | 2023-04-26 02:19:00 |
| refinenet_HPM | refinenet_HPM | SJTU | gzx | refinenet_HPM | 2019-03-01 09:22:09 |
| unet_resnet50 | unet_resnet50 | hainnu | nothing | unet_resnet50 | 2023-08-13 16:08:21 |
| vgg_unet | vgg_unet | hainnu | golangboy | nothing | 2023-08-13 12:13:17 |
| weakly_seg_validation_test | weakly_seg_validation_test | NEU | Smile Lab | weakly_seg_validation_test | 2019-09-08 01:18:19 |
| weight+RS | weight+RS | Northwestern Politechnical University | Peng Wang, Shunhua Shen | weight+RS | 2019-03-24 08:13:07 |