cellmap_organelle_semantic
The artifact/checkpoint/view links below point to locations on the Janelia cluster and will only work on the Janelia network.
Views: neuroglancer views for every row below
Region:
- jrc_cos7-1a_crop236 400x400x400
- jrc_cos7-1a_crop243 400x400x400
- jrc_cos7-1b_crop235 400x400x400
- jrc_cos7-1b_crop241 400x400x400
- jrc_ctl-id8-1_crop119 400x400x400
- jrc_fly-mb-1a_crop123 400x400x400
- jrc_fly-vnc-1_crop174 400x400x400
- jrc_hela-2_crop14 130x300x300
- jrc_hela-2_crop23 500x500x500
- jrc_hela-2_crop3 500x800x800
- jrc_hela-2_crop57 400x400x400
- jrc_hela-2_crop9 106x200x200
- jrc_hela-3_crop33 400x400x400
- jrc_hela-3_crop34 400x400x400
- jrc_hela-3_crop61 400x400x400
- jrc_jurkat-1_crop112 1000x500x1000
- jrc_jurkat-1_crop66 400x400x400
- jrc_jurkat-1_crop68 400x400x400
- jrc_macrophage-2_crop39 400x400x400
- jrc_macrophage-2_crop40 400x400x400
- jrc_macrophage-2_crop73 400x400x400
- jrc_mus-kidney_crop129 300x300x300
- jrc_mus-kidney_crop159 200x200x200
- jrc_mus-kidney_crop161 200x200x200
- jrc_mus-liver-zon-1_crop267 400x400x400
- jrc_mus-liver-zon-1_crop274 400x400x400
- jrc_mus-liver-zon-1_crop279 400x400x400
- jrc_mus-liver-zon-1_crop313 400x400x400
- jrc_mus-liver-zon-1_crop323 400x400x400
- jrc_mus-liver-zon-2_crop342 400x400x400
- jrc_mus-liver-zon-2_crop370 400x400x400
- jrc_mus-liver_crop138 400x400x400
- jrc_mus-liver_crop139 200x200x200
- jrc_mus-liver_crop145 360x750x750
- jrc_sum159-1_crop20 170x400x400
- jrc_sum159-1_crop81 400x400x400
- jrc_sum159-4_crop188 400x400x400
- jrc_sum159-4_crop209 400x400x400
- jrc_sum159-4_crop217 400x400x400
- jrc_ut21-1413-003_crop191 400x400x400
- jrc_ut21-1413-003_crop224 400x400x400
- jrc_ut21-1413-003_crop225 400x400x400
- jrc_ut21-1413-003_crop226 400x400x400
| # | model | links | semantic.mean_iou (higher is better) | postprocess | semantic.mean_dice | semantic.pixel_accuracy | semantic.classes_present |
|---|---|---|---|---|---|---|---|
| 1 | lmd2cm__1c_simmim_lvd_20261028_211227.step25000.identity | artifactscheckpoint | 0.4151 | Identity | 0.5271 | 0.8898 | 48.0000 |
| 2 | lmd2cm__1c_simmim_lvd_20261028_211227.step50000.identity | artifactscheckpoint | 0.3984 | Identity | 0.5170 | 0.8725 | 48.0000 |
| 3 | lmd2cm__2c_affinity_simmim_lvd_seed1_20261031_013819.step25000.identity | artifactscheckpoint | 0.3900 | Identity | 0.4951 | 0.8779 | 48.0000 |
| 4 | lmd2cm__1b_lvd_20261028_204534.step25000.identity | artifactscheckpoint | 0.3688 | Identity | 0.4795 | 0.8755 | 48.0000 |
| 5 | lmd2cm__2a_affinity_simmim_lvd_native_20261029_221401.step25000.identity | artifactscheckpoint | 0.3658 | Identity | 0.4849 | 0.8447 | 48.0000 |
| 6 | lmd2cm__1e_affinity_simmim_lvd_20261029_204759.step25000.identity | artifactscheckpoint | 0.3347 | Identity | 0.4389 | 0.8557 | 48.0000 |
| 7 | lmd2cm__2a_affinity_simmim_lvd_native_20261029_221401.step50000.identity | artifactscheckpoint | 0.3324 | Identity | 0.4445 | 0.8015 | 48.0000 |
| 8 | lmd2cm__2b_affinity_simmim_lvd_dice_20261030_165448.step25000.identity | artifactscheckpoint | 0.3277 | Identity | 0.4423 | 0.8183 | 48.0000 |
| 9 | lmd2cm__1c_simmim_lvd_20261028_211227.step75000.identity | artifactscheckpoint | 0.3265 | Identity | 0.4377 | 0.8207 | 48.0000 |
| 10 | lmd2cm__1d_simmim_scratch_20261029_144643.step50000.identity | artifactscheckpoint | 0.3238 | Identity | 0.4289 | 0.8504 | 48.0000 |
| 11 | lmd2cm__1d_simmim_scratch_20261029_144643.step25000.identity | artifactscheckpoint | 0.3227 | Identity | 0.4193 | 0.8471 | 48.0000 |
| 12 | lmd2cm__2c_affinity_simmim_lvd_seed1_20261031_013819.step50000.identity | artifactscheckpoint | 0.3214 | Identity | 0.4327 | 0.8204 | 48.0000 |
| 13 | lmd2cm__2a_affinity_simmim_lvd_native_20261029_221401.step75000.identity | artifactscheckpoint | 0.3100 | Identity | 0.4193 | 0.7437 | 48.0000 |
| 14 | lmd2cm__1b_lvd_20261028_204534.step50000.identity | artifactscheckpoint | 0.2997 | Identity | 0.4017 | 0.8272 | 48.0000 |
| 15 | lmd2cm__1c_simmim_lvd_20261028_211227.step100000.identity | artifactscheckpoint | 0.2840 | Identity | 0.3856 | 0.7858 | 48.0000 |
| 16 | lmd2cm__1d_simmim_scratch_20261029_144643.step75000.identity | artifactscheckpoint | 0.2665 | Identity | 0.3702 | 0.7791 | 48.0000 |
| 17 | lmd2cm__2a_affinity_simmim_lvd_native_20261029_221401.step100000.identity | artifactscheckpoint | 0.2509 | Identity | 0.3458 | 0.6106 | 48.0000 |
| 18 | lmd2cm__1d_simmim_scratch_20261029_144643.step100000.identity | artifactscheckpoint | 0.2466 | Identity | 0.3503 | 0.7431 | 48.0000 |
| 19 | lmd2cm__1e_affinity_simmim_lvd_20261029_204759.step50000.identity | artifactscheckpoint | 0.2319 | Identity | 0.3193 | 0.6873 | 48.0000 |
| 20 | lmd2cm__2b_affinity_simmim_lvd_dice_20261030_165448.step50000.identity | artifactscheckpoint | 0.2261 | Identity | 0.3258 | 0.6540 | 48.0000 |
| 21 | lmd2cm__2c_affinity_simmim_lvd_seed1_20261031_013819.step75000.identity | artifactscheckpoint | 0.2241 | Identity | 0.3249 | 0.6943 | 48.0000 |
| 22 | lmd2cm__1b_lvd_20261028_204534.step75000.identity | artifactscheckpoint | 0.2174 | Identity | 0.3095 | 0.7034 | 48.0000 |
| 23 | lmd2cm__1b_lvd_20261028_204534.step100000.identity | artifactscheckpoint | 0.1820 | Identity | 0.2608 | 0.6494 | 48.0000 |
| 24 | lmd2cm__2b_affinity_simmim_lvd_dice_20261030_165448.step75000.identity | artifactscheckpoint | 0.1816 | Identity | 0.2660 | 0.5663 | 48.0000 |
| 25 | lmd2cm__1a_scratch_20261028_192039.step100000.identity | artifactscheckpoint | 0.1794 | Identity | 0.2549 | 0.7132 | 48.0000 |
| 26 | lmd2cm__2c_affinity_simmim_lvd_seed1_20261031_013819.step100000.identity | artifactscheckpoint | 0.1776 | Identity | 0.2632 | 0.5914 | 48.0000 |
| 27 | lmd2cm__1a_scratch_20261028_192039.step75000.identity | artifactscheckpoint | 0.1548 | Identity | 0.2221 | 0.6977 | 48.0000 |
| 28 | lmd2cm__1e_affinity_simmim_lvd_20261029_204759.step100000.identity | artifactscheckpoint | 0.1453 | Identity | 0.2089 | 0.5379 | 48.0000 |
| 29 | lmd2cm__2b_affinity_simmim_lvd_dice_20261030_165448.step100000.identity | artifactscheckpoint | 0.1408 | Identity | 0.2070 | 0.4882 | 48.0000 |
| 30 | lmd2cm__1e_affinity_simmim_lvd_20261029_204759.step75000.identity | artifactscheckpoint | 0.1348 | Identity | 0.2003 | 0.4819 | 48.0000 |
| 31 | lmd2cm__1a_scratch_20261028_192039.step50000.identity | artifactscheckpoint | 0.1139 | Identity | 0.1701 | 0.6098 | 48.0000 |
| 32 | lmd2cm__1a_scratch_20261028_192039.step25000.identity | artifactscheckpoint | 0.0717 | Identity | 0.1026 | 0.6123 | 48.0000 |
- ↑ mean_iou
- Mean intersection-over-union over the classes present.
- ↑ mean_dice
- Mean Dice coefficient over the classes present.
- ↑ pixel_accuracy
- Fraction of voxels given the correct class.
- classes_present
- Number of classes present in the ground truth.
gary_comparison_neuron_instance
The artifact/checkpoint/view links below point to locations on the Janelia cluster and will only work on the Janelia network.
Views: neuroglancer views for every row below
Region:
- hemibrain_eb_test 896x896x896 (sub-region)
| # | model | links | voxel_instance.pq (higher is better) | postprocess | voxel_instance.voi_merge | voxel_instance.voi_split | voxel_instance.sq | voxel_instance.rq | voxel_instance.adapted_rand_error |
|---|---|---|---|---|---|---|---|---|---|
| 1 | gary__4b_dinov3sat_axial_subpixel_p8_20260923_173431.step450000.mws_size_filter | artifactscheckpoint | 0.1711 | mws(repulsive_stride=1, min_size=20000) | 0.3870 | 0.9332 | 0.8560 | 0.1999 | 0.2174 |
| 2 | gary__5a_dinov3_axial_unetr_20260924_001902.step500000.mws_size_filter | artifactscheckpoint | 0.1682 | mws(repulsive_stride=1, min_size=20000) | 0.3981 | 0.9115 | 0.8498 | 0.1980 | 0.2061 |
| 3 | gary__1c_dinov3_axial_subpixel_1m_20260919_124840.step1000000.mws_size_filter | artifactscheckpoint | 0.1665 | mws(repulsive_stride=1, min_size=20000) | 0.4330 | 0.9238 | 0.8542 | 0.1949 | 0.2165 |
| 4 | gary__1a_dinov3_axial_subpixel_20260916_215544.step500000.mws_size_filter | artifactscheckpoint | 0.1636 | mws(repulsive_stride=1, min_size=20000) | 0.4209 | 0.9544 | 0.8463 | 0.1934 | 0.2219 |
| 5 | gary__3b_simmim85_axial_subpixel_20260922_163950.step500000.mws_size_filter | artifactscheckpoint | 0.1608 | mws(repulsive_stride=1, min_size=20000) | 0.4394 | 0.9578 | 0.8526 | 0.1886 | 0.2287 |
| 6 | gary__1a_dinov3_axial_subpixel_20260916_215544.step300000.mws_size_filter | artifactscheckpoint | 0.1570 | mws(repulsive_stride=1, min_size=20000) | 0.4554 | 0.9707 | 0.8458 | 0.1856 | 0.2288 |
| 7 | gary__6a_dinov3_axial_subpixel_erode1_20260926_212744.step200000.mws_size_filter_fill | artifactscheckpoint | 0.1569 | mws(repulsive_stride=1, min_size=20000, fill=1) | 0.5094 | 1.0259 | 0.8463 | 0.1853 | 0.2536 |
| 8 | gary__6a_dinov3_axial_subpixel_erode1_20260926_212744.step500000.mws_size_filter_fill | artifactscheckpoint | 0.1560 | mws(repulsive_stride=1, min_size=20000, fill=1) | 0.4698 | 0.9968 | 0.8500 | 0.1836 | 0.2399 |
| 9 | gary__3d_simmim90_hemibrain_500k_axial_subpixel_20260928_185741.step200000.mws_size_filter | artifactscheckpoint | 0.1553 | mws(repulsive_stride=1, min_size=20000) | 0.4859 | 0.9677 | 0.8478 | 0.1832 | 0.2377 |
| 10 | gary__1b_scratch_axial_subpixel_20260917_093923.step500000.mws_size_filter | artifactscheckpoint | 0.1531 | mws(repulsive_stride=1, min_size=20000) | 0.4800 | 0.9646 | 0.8560 | 0.1789 | 0.2370 |
| 11 | gary__1b_scratch_axial_subpixel_20260917_093923.step300000.mws_size_filter | artifactscheckpoint | 0.1511 | mws(repulsive_stride=1, min_size=20000) | 0.5119 | 0.9631 | 0.8488 | 0.1780 | 0.2376 |
| 12 | gary__1a_dinov3_axial_subpixel_20260916_215544.step100000.mws_size_filter | artifactscheckpoint | 0.1489 | mws(repulsive_stride=1, min_size=20000) | 0.5307 | 0.9718 | 0.8490 | 0.1753 | 0.2389 |
| 13 | gary__6a_dinov3_axial_subpixel_erode1_20260926_212744.step500000.mws_size_filter | artifactscheckpoint | 0.1445 | mws(repulsive_stride=1, min_size=20000) | 0.7902 | 1.0787 | 0.8072 | 0.1790 | 0.4039 |
| 14 | gary__6a_dinov3_axial_subpixel_erode1_20260926_212744.step200000.mws_size_filter | artifactscheckpoint | 0.1433 | mws(repulsive_stride=1, min_size=20000) | 0.8332 | 1.1005 | 0.8080 | 0.1774 | 0.4249 |
| 15 | gary__1b_scratch_axial_subpixel_20260917_093923.step100000.mws_size_filter | artifactscheckpoint | 0.1400 | mws(repulsive_stride=1, min_size=20000) | 0.6622 | 0.9803 | 0.8367 | 0.1673 | 0.2465 |
| 16 | gary__1c_dinov3_axial_subpixel_1m_20260919_124840.step1000000.cc_threshold_size_filter | artifactscheckpoint | 0.1295 | cc_threshold(logit=+3, thr=0.6457, min_size=500) | 3.3158 | 0.7559 | 0.7491 | 0.1728 | 0.8852 |
| 17 | gary__1a_dinov3_axial_subpixel_20260916_215544.step500000.cc_threshold_size_filter | artifactscheckpoint | 0.1184 | cc_threshold(logit=+6, thr=0.7685, min_size=5000) | 1.5251 | 1.2043 | 0.7603 | 0.1558 | 0.7282 |
| 18 | gary__3b_simmim85_axial_subpixel_20260922_163950.step500000.cc_threshold_size_filter | artifactscheckpoint | 0.1162 | cc_threshold(logit=+6, thr=0.7685, min_size=5000) | 1.5500 | 1.1894 | 0.7588 | 0.1531 | 0.7279 |
| 19 | gary__1b_scratch_axial_subpixel_20260917_093923.step500000.cc_threshold_size_filter | artifactscheckpoint | 0.1137 | cc_threshold(logit=+6, thr=0.7685, min_size=5000) | 1.6186 | 1.2073 | 0.7496 | 0.1517 | 0.7479 |
| 20 | gary__1a_dinov3_axial_subpixel_20260916_215544.step300000.cc_threshold_size_filter | artifactscheckpoint | 0.1116 | cc_threshold(logit=+6, thr=0.7685, min_size=5000) | 1.6546 | 1.2127 | 0.7407 | 0.1507 | 0.7552 |
| 21 | gary__1a_dinov3_axial_subpixel_20260916_215544.step100000.cc_threshold_size_filter | artifactscheckpoint | 0.1098 | cc_threshold(logit=+6, thr=0.7685, min_size=5000) | 1.7038 | 1.2250 | 0.7371 | 0.1490 | 0.7683 |
| 22 | gary__1b_scratch_axial_subpixel_20260917_093923.step300000.cc_threshold_size_filter | artifactscheckpoint | 0.1008 | cc_threshold(logit=+6, thr=0.7685, min_size=5000) | 1.8807 | 1.2916 | 0.7238 | 0.1393 | 0.8159 |
| 23 | gary__1b_scratch_axial_subpixel_20260917_093923.step100000.cc_threshold_size_filter | artifactscheckpoint | 0.0806 | cc_threshold(logit=+6, thr=0.7685, min_size=20000) | 2.2205 | 1.2962 | 0.6814 | 0.1182 | 0.8518 |
- ↑ pq
- Panoptic quality, sq * rq: how well predicted objects match true ones at IoU > 0.5.
- ↓ voi_merge
- Variation of information, merge term H(truth | prediction): under-segmentation.
- ↓ voi_split
- Variation of information, split term H(prediction | truth): over-segmentation.
- ↑ sq
- Segmentation quality: mean IoU of the matched object pairs.
- ↑ rq
- Recognition quality: F1 of object matching, balancing missed and spurious objects.
- ↓ adapted_rand_error
- Adapted Rand error, 1 - Rand F-score of the voxel pairing.
hemibrain_eb_neurite_tracing
The artifact/checkpoint/view links below point to locations on the Janelia cluster and will only work on the Janelia network.
Views: neuroglancer views for every row below
Region:
- hemibrain_eb_roi_1 1475x1475x1475
- hemibrain_eb_roi_2 2725x2725x2725
- hemibrain_eb_roi_3 2100x2100x2100
| # | model | links | skeleton_erl.nerl (higher is better) | postprocess | skeleton_erl.erl_um | skeleton_erl.voi_sum | skeleton_erl.voi_split | skeleton_erl.voi_merge | skeleton_erl.n_non0_mergers | skeleton_erl.n_splits |
|---|---|---|---|---|---|---|---|---|---|---|
| 1 | ffn_januszewski2018.identity | artifacts | 0.8253 | Identity | 64.8843 | 0.2803 | 0.2475 | 0.0328 | 13.6667 | 125.0000 |
- ↑ nerl
- Normalised expected run length: ERL along the traced skeletons divided by the maximum possible.
- ↑ erl_um
- Expected run length in micrometres.
- ↓ voi_sum
- voi_split + voi_merge.
- ↓ voi_split
- Variation of information on the skeleton nodes, split term H(prediction | truth): over-segmentation.
- ↓ voi_merge
- Variation of information on the skeleton nodes, merge term H(truth | prediction): under-segmentation.
- ↓ n_non0_mergers
- Number of (merging segment, skeleton) pairs where one segment spans several skeletons.
- ↓ n_splits
- Number of skeleton edges the segmentation cuts.
lm_zebrafish_cell_instance
The artifact/checkpoint/view links below point to locations on the Janelia cluster and will only work on the Janelia network.
Views: neuroglancer views for every row below
Region:
- lm_zebrafish_t10 152x508x1466
| # | model | links | voxel_instance.pq (higher is better) | postprocess | voxel_instance.voi_merge | voxel_instance.voi_split | voxel_instance.sq | voxel_instance.rq | voxel_instance.adapted_rand_error |
|---|---|---|---|---|---|---|---|---|---|
| 1 | cellpose-sam_ft_t1-t25.step0.identity | artifacts | 0.6275 | Identity | 0.4869 | 0.5966 | 0.8353 | 0.7512 | 0.2124 |
| 2 | cellpose-sam_ft_t25.step0.identity | artifacts | 0.6265 | Identity | 0.4592 | 0.6161 | 0.8277 | 0.7569 | 0.1908 |
| 3 | cellpose-sam_ft_t1.step0.identity | artifacts | 0.6012 | Identity | 0.5786 | 0.6361 | 0.8395 | 0.7162 | 0.2671 |
| 4 | lsd-deep4wide_ft_t1-t25.step0.identity_size_filter_fill | artifacts | 0.5918 | identity(min_size=50000, fill_distance=3) | 0.9333 | 0.7356 | 0.8177 | 0.7237 | 0.4362 |
| 5 | lsd-deep4wide_ft_t1-t25.step0.identity_size_filter | artifacts | 0.5274 | identity(min_size=75000) | 2.2114 | 0.8567 | 0.7469 | 0.7061 | 0.9137 |
| 6 | cellpose-sam_zeroshot.step0.identity | artifacts | 0.4952 | Identity | 1.0795 | 0.9393 | 0.7756 | 0.6385 | 0.4797 |
- ↑ pq
- Panoptic quality, sq * rq: how well predicted objects match true ones at IoU > 0.5.
- ↓ voi_merge
- Variation of information, merge term H(truth | prediction): under-segmentation.
- ↓ voi_split
- Variation of information, split term H(prediction | truth): over-segmentation.
- ↑ sq
- Segmentation quality: mean IoU of the matched object pairs.
- ↑ rq
- Recognition quality: F1 of object matching, balancing missed and spurious objects.
- ↓ adapted_rand_error
- Adapted Rand error, 1 - Rand F-score of the voxel pairing.
lmd_ssl_v1_neuron_instance
The artifact/checkpoint/view links below point to locations on the Janelia cluster and will only work on the Janelia network.
Views: neuroglancer views for every row below
Region:
- kasthuri15_ac4 256x640x640 (sub-region)
- liconn_expid82 768x1152x1152 (sub-region)
- liconn_mouse_hippocampus 512x512x512 (sub-region)
- zebrafish_fish2_doublecube1 1920x1920x1920 (sub-region)
| # | model | links | voxel_instance.pq (higher is better) | postprocess | voxel_instance.voi_merge | voxel_instance.voi_split | voxel_instance.sq | voxel_instance.rq | voxel_instance.adapted_rand_error |
|---|---|---|---|---|---|---|---|---|---|
| 1 | sam1__arm8_8nm_gb16_musam128_r0_20260916_223719.step200000.identity_size_filter | artifactscheckpoint | 0.2786 | identity(min_size=5000) | 2.0173 | 1.1858 | 0.6867 | 0.3951 | 0.5714 |
| 2 | sam1__arm12_8nm_gb16_musam128_sat493m_r0_20260923_110837.step200000.identity_size_filter | artifactscheckpoint | 0.2775 | identity(min_size=5000) | 2.2002 | 1.0969 | 0.6903 | 0.3880 | 0.5500 |
| 3 | sam1__arm7_8nm_gb16_musam64_r0_20260916_105148.step200000.identity_size_filter | artifactscheckpoint | 0.2591 | identity(min_size=5000) | 3.3298 | 0.8670 | 0.6893 | 0.3615 | 0.6418 |
| 4 | lmd1__1c_dinov3_simmim_ft_subpixel_20260827_231846.step50000.identity_size_filter | artifactscheckpoint | 0.2369 | identity(min_size=50000) | 2.0755 | 1.3125 | 0.7078 | 0.3241 | 0.5426 |
| 5 | lmd1__2c_dinov3_lvd_ft_subpixel_20260824_183919.step50000.identity_size_filter | artifactscheckpoint | 0.2287 | identity(min_size=50000) | 1.8565 | 1.3023 | 0.7072 | 0.3155 | 0.5380 |
| 6 | sam1__arm11_8nm_gb16_musam128_scratch_r0_20260923_110902.step200000.identity_size_filter | artifactscheckpoint | 0.2254 | identity(min_size=5000) | 3.6099 | 0.8798 | 0.6947 | 0.3129 | 0.6648 |
| 7 | sam1__arm6_8nm_gb16_musam_r0_20260915_124216.step200000.identity_size_filter | artifactscheckpoint | 0.2244 | identity(min_size=5000) | 3.5374 | 0.9153 | 0.6972 | 0.3111 | 0.6831 |
| 8 | sam1__arm4_8nm_gb16_r0_20260913_221637.step200000.identity_size_filter | artifactscheckpoint | 0.2094 | identity(min_size=5000) | 3.4492 | 0.9709 | 0.6909 | 0.2945 | 0.6829 |
| 9 | sam1__arm5_8nm_gb32_r0_20260913_223406.step200000.identity_size_filter | artifactscheckpoint | 0.2038 | identity(min_size=5000) | 3.9397 | 0.7063 | 0.7221 | 0.2775 | 0.6845 |
| 10 | sam1__arm9_8nm_gb64_musam32_c128_r0_20260921_004737.step200000.identity_size_filter | artifactscheckpoint | 0.2017 | identity(min_size=5000) | 3.3340 | 0.9779 | 0.6943 | 0.2821 | 0.6261 |
| 11 | sam1__arm3_p8_r0_20260913_200809.step200000.identity_size_filter | artifactscheckpoint | 0.1734 | identity(min_size=5000) | 4.1616 | 0.9427 | 0.7112 | 0.2391 | 0.7769 |
| 12 | lmd1__2c_dinov3_lvd_ft_subpixel_20260824_183919.step50000.cc_threshold_size_filter | artifactscheckpoint | 0.1420 | cc_threshold(logit=+3, thr=0.6457, min_size=50000) | 3.7074 | 1.0675 | 0.6507 | 0.2110 | 0.7435 |
| 13 | lmd1__2c_dinov3_lvd_ft_subpixel_20260824_183919.step50000.cc_threshold | artifactscheckpoint | 0.0031 | cc_threshold(logit=+0, thr=0.5000) | 6.7441 | 0.4306 | 0.5048 | 0.0044 | 0.8968 |
- ↑ pq
- Panoptic quality, sq * rq: how well predicted objects match true ones at IoU > 0.5.
- ↓ voi_merge
- Variation of information, merge term H(truth | prediction): under-segmentation.
- ↓ voi_split
- Variation of information, split term H(prediction | truth): over-segmentation.
- ↑ sq
- Segmentation quality: mean IoU of the matched object pairs.
- ↑ rq
- Recognition quality: F1 of object matching, balancing missed and spurious objects.
- ↓ adapted_rand_error
- Adapted Rand error, 1 - Rand F-score of the voxel pairing.
lmd_ssl_v1_zebrafish_instance
The artifact/checkpoint/view links below point to locations on the Janelia cluster and will only work on the Janelia network.
Views: neuroglancer views for every row below
Region:
- zebrafish_fish2_doublecube1 1920x1920x1920 (sub-region)
| # | model | links | voxel_instance.pq (higher is better) | postprocess | voxel_instance.voi_merge | voxel_instance.voi_split | voxel_instance.sq | voxel_instance.rq | voxel_instance.adapted_rand_error |
|---|---|---|---|---|---|---|---|---|---|
| 1 | lmd1__1c_dinov3_simmim_ft_subpixel_20260827_231846.step50000.identity_size_filter | artifactscheckpoint | 0.0876 | identity(min_size=50000) | 3.6554 | 2.0653 | 0.7173 | 0.1221 | 0.9370 |
| 2 | lmd1__2c_dinov3_lvd_ft_subpixel_20260824_183919.step50000.identity_size_filter | artifactscheckpoint | 0.0862 | identity(min_size=50000) | 3.1383 | 2.0274 | 0.6978 | 0.1236 | 0.9317 |
| 3 | lmd1__3c_muvit_mae_ft_subpixel_20260905_074206.step50000.identity_size_filter | artifactscheckpoint | 0.0719 | identity(min_size=50000) | 4.6463 | 2.2950 | 0.7400 | 0.0972 | 0.9593 |
- ↑ pq
- Panoptic quality, sq * rq: how well predicted objects match true ones at IoU > 0.5.
- ↓ voi_merge
- Variation of information, merge term H(truth | prediction): under-segmentation.
- ↓ voi_split
- Variation of information, split term H(prediction | truth): over-segmentation.
- ↑ sq
- Segmentation quality: mean IoU of the matched object pairs.
- ↑ rq
- Recognition quality: F1 of object matching, balancing missed and spurious objects.
- ↓ adapted_rand_error
- Adapted Rand error, 1 - Rand F-score of the voxel pairing.
lmd_ssl_v2_neuron_instance
The artifact/checkpoint/view links below point to locations on the Janelia cluster and will only work on the Janelia network.
Views: neuroglancer views for every row below
Region:
- cns_mito_crop164 1152x1152x1792
- cremi_sample_c 512x512x512 (sub-region)
- flyid49_40xw006 640x256x256 (sub-region)
- kasthuri15_ac4 256x640x640 (sub-region)
- liconn_expid82 768x1152x1152 (sub-region)
- liconn_mouse_hippocampus 512x512x512 (sub-region)
- zebrafish_fish2_doublecube1 1920x1920x1920 (sub-region)
| # | model | links | voxel_instance.pq (higher is better) | postprocess | voxel_instance.voi_merge | voxel_instance.voi_split | voxel_instance.sq | voxel_instance.rq | voxel_instance.adapted_rand_error |
|---|---|---|---|---|---|---|---|---|---|
| 1 | lmd2__2f_affinity_unetr_simmim_lvd_20261020_200659.step100000.mws_size_filter | artifactscheckpoint | 0.2771 | mws(repulsive_stride=1, min_size=20000) | 1.1534 | 0.9145 | 0.7463 | 0.3630 | 0.3370 |
| 2 | lmd2__2d_sam_musam128_lvd_20261018_231122.step100000.identity_size_filter | artifactscheckpoint | 0.2663 | identity(min_size=5000) | 1.8480 | 1.0293 | 0.7019 | 0.3710 | 0.5322 |
| 3 | lmd2__2h_affinity_unetr_simmim_lvd_mask60_g2_20261021_071125.step100000.mws_size_filter | artifactscheckpoint | 0.2640 | mws(repulsive_stride=1, min_size=20000) | 1.1869 | 0.9412 | 0.7479 | 0.3449 | 0.3534 |
| 4 | lmd2__2e_mtlsd_unetr_lvd_w10_20261019_004951.step100000.mws_size_filter | artifactscheckpoint | 0.2551 | mws(repulsive_stride=1, min_size=20000) | 1.4273 | 1.0329 | 0.7435 | 0.3313 | 0.4154 |
| 5 | lmd2__2a_affinity_unetr_lvd_20261018_215936.step100000.mws_size_filter | artifactscheckpoint | 0.2455 | mws(repulsive_stride=1, min_size=20000) | 1.3620 | 1.0264 | 0.7434 | 0.3211 | 0.3894 |
| 6 | lmd2__2g_affinity_unetr_simmim_scratch_20261020_124554.step100000.mws_size_filter | artifactscheckpoint | 0.2410 | mws(repulsive_stride=1, min_size=20000) | 1.2720 | 1.0544 | 0.7415 | 0.3162 | 0.3834 |
| 7 | lmd2__2b_affinity_unetr_scratch_20261018_215937.step100000.mws_size_filter | artifactscheckpoint | 0.2240 | mws(repulsive_stride=1, min_size=20000) | 1.6018 | 1.1764 | 0.7336 | 0.3002 | 0.4938 |
| 8 | lmd2__2c_mtlsd_unetr_lvd_20261018_215102.step100000.mws_size_filter | artifactscheckpoint | 0.2043 | mws(repulsive_stride=1, min_size=20000) | 1.6373 | 1.2736 | 0.7265 | 0.2640 | 0.4526 |
- ↑ pq
- Panoptic quality, sq * rq: how well predicted objects match true ones at IoU > 0.5.
- ↓ voi_merge
- Variation of information, merge term H(truth | prediction): under-segmentation.
- ↓ voi_split
- Variation of information, split term H(prediction | truth): over-segmentation.
- ↑ sq
- Segmentation quality: mean IoU of the matched object pairs.
- ↑ rq
- Recognition quality: F1 of object matching, balancing missed and spurious objects.
- ↓ adapted_rand_error
- Adapted Rand error, 1 - Rand F-score of the voxel pairing.
lmd_ssl_v2_neuron_instance_crop512
The artifact/checkpoint/view links below point to locations on the Janelia cluster and will only work on the Janelia network.
Views: neuroglancer views for every row below
Region:
- cns_mito_crop164_crop512 512x512x512
- cremi_sample_c_crop512 512x256x256 (sub-region)
- flyid49_40xw006_crop512 640x256x256 (sub-region)
- kasthuri15_ac4_crop512 256x256x256 (sub-region)
- liconn_expid82_crop512 768x512x512 (sub-region)
- liconn_mouse_hippocampus_crop512 512x512x512 (sub-region)
- zebrafish_fish2_doublecube1_crop512 512x512x384 (sub-region)
| # | model | links | voxel_instance.pq (higher is better) | postprocess | voxel_instance.voi_merge | voxel_instance.voi_split | voxel_instance.sq | voxel_instance.rq | voxel_instance.adapted_rand_error |
|---|---|---|---|---|---|---|---|---|---|
| 1 | lmd2__2f_affinity_unetr_simmim_lvd_20261020_200659.step100000.mws_size_filter | artifactscheckpoint | 0.2998 | mws(repulsive_stride=1, min_size=20000) | 1.0813 | 0.7477 | 0.7523 | 0.3930 | 0.2974 |
| 2 | lmd2__2d_sam_musam128_lvd_20261018_231122.step75000.identity_size_filter | artifactscheckpoint | 0.2933 | identity(min_size=500) | 1.6466 | 0.8547 | 0.7108 | 0.4064 | 0.5013 |
| 3 | lmd2__2h_affinity_unetr_simmim_lvd_mask60_g2_20261021_071125.step100000.mws_size_filter | artifactscheckpoint | 0.2891 | mws(repulsive_stride=1, min_size=20000) | 1.0867 | 0.7303 | 0.7613 | 0.3745 | 0.3013 |
| 4 | lmd2__2g_affinity_unetr_simmim_scratch_20261020_124554.step75000.mws_size_filter | artifactscheckpoint | 0.2811 | mws(repulsive_stride=1, min_size=20000) | 1.1369 | 0.8312 | 0.7480 | 0.3689 | 0.3265 |
| 5 | lmd2__2e_mtlsd_unetr_lvd_w10_20261019_004951.step100000.mws_size_filter | artifactscheckpoint | 0.2767 | mws(repulsive_stride=1, min_size=20000) | 1.2545 | 0.7881 | 0.7506 | 0.3585 | 0.3263 |
| 6 | lmd2__2a_affinity_unetr_lvd_20261018_215936.step100000.mws_size_filter | artifactscheckpoint | 0.2711 | mws(repulsive_stride=1, min_size=20000) | 1.2243 | 0.8179 | 0.7527 | 0.3526 | 0.3347 |
| 7 | lmd2__2b_affinity_unetr_scratch_20261018_215937.step100000.mws_size_filter | artifactscheckpoint | 0.2508 | mws(repulsive_stride=1, min_size=20000) | 1.4137 | 0.9018 | 0.7360 | 0.3360 | 0.4033 |
| 8 | lmd2__2c_mtlsd_unetr_lvd_20261018_215102.step100000.mws_size_filter | artifactscheckpoint | 0.2185 | mws(repulsive_stride=1, min_size=20000) | 1.4994 | 1.0359 | 0.7297 | 0.2809 | 0.4160 |
- ↑ pq
- Panoptic quality, sq * rq: how well predicted objects match true ones at IoU > 0.5.
- ↓ voi_merge
- Variation of information, merge term H(truth | prediction): under-segmentation.
- ↓ voi_split
- Variation of information, split term H(prediction | truth): over-segmentation.
- ↑ sq
- Segmentation quality: mean IoU of the matched object pairs.
- ↑ rq
- Recognition quality: F1 of object matching, balancing missed and spurious objects.
- ↓ adapted_rand_error
- Adapted Rand error, 1 - Rand F-score of the voxel pairing.
nisb_base_neurite_tracing
The artifact/checkpoint/view links below point to locations on the Janelia cluster and will only work on the Janelia network.
Views: neuroglancer views for every row below
Region:
- nisb_base_seed101 3000x3000x1350
| # | model | links | skeleton_erl.nerl (higher is better) | postprocess | skeleton_erl.erl_um | skeleton_erl.voi_sum | skeleton_erl.voi_split | skeleton_erl.voi_merge | skeleton_erl.n_non0_mergers | skeleton_erl.n_splits |
|---|---|---|---|---|---|---|---|---|---|---|
| 1 | large_inputs_nisb__w512_5cube_20261003_141342.step500000.cc_threshold | artifactscheckpoint | 0.5639 | cc_threshold(logit=+7, thr=0.8022) | 144.6894 | 2.0205 | 1.3375 | 0.6829 | 79.0000 | 3417.0000 |
| 2 | large_inputs_nisb__w512_5cube_20261003_141342.step300000.cc_threshold | artifacts (missing)checkpoint | 0.5453 | cc_threshold(logit=+6, thr=0.7685) | 139.9030 | 1.9672 | 1.3020 | 0.6652 | 98.0000 | 3262.0000 |
| 3 | large_inputs_nisb__w512_5cube_20261003_141342.step400000.cc_threshold | artifacts (missing)checkpoint | 0.5364 | cc_threshold(logit=+6, thr=0.7685) | 137.6168 | 1.8584 | 1.2668 | 0.5917 | 138.0000 | 3181.0000 |
| 4 | large_inputs_nisb__w512_5cube_20261003_141342.step200000.cc_threshold | artifacts (missing)checkpoint | 0.5296 | cc_threshold(logit=+6, thr=0.7685) | 135.8933 | 2.1970 | 1.3886 | 0.8084 | 74.0000 | 3324.0000 |
| 5 | large_inputs_nisb__w512_3cube_20261003_141341.step200000.cc_threshold | artifacts (missing)checkpoint | 0.5125 | cc_threshold(logit=+7, thr=0.8022) | 131.4910 | 2.3957 | 1.4550 | 0.9407 | 79.0000 | 3560.0000 |
| 6 | large_inputs_nisb__w512_3cube_20261003_141341.step300000.cc_threshold | artifacts (missing)checkpoint | 0.4879 | cc_threshold(logit=+8, thr=0.8320) | 125.1929 | 2.3083 | 1.4413 | 0.8671 | 138.0000 | 4007.0000 |
| 7 | large_inputs_nisb__w256_5cube_20261003_141051.step500000.cc_threshold | artifacts (missing)checkpoint | 0.4840 | cc_threshold(logit=+6, thr=0.7685) | 124.1810 | 2.3547 | 1.4482 | 0.9066 | 121.0000 | 2787.0000 |
| 8 | large_inputs_nisb__w512_3cube_20261003_141341.step100000.cc_threshold | artifacts (missing)checkpoint | 0.4703 | cc_threshold(logit=+6, thr=0.7685) | 120.6568 | 2.5000 | 1.4770 | 1.0230 | 109.0000 | 3548.0000 |
| 9 | large_inputs_nisb__w512_5cube_20261003_141342.step100000.cc_threshold | artifacts (missing)checkpoint | 0.4692 | cc_threshold(logit=+6, thr=0.7685) | 120.3784 | 2.4630 | 1.4859 | 0.9770 | 121.0000 | 3770.0000 |
| 10 | large_inputs_nisb__w256_1cube_20261003_141051.step500000.cc_threshold | artifacts (missing)checkpoint | 0.4654 | cc_threshold(logit=+7, thr=0.8022) | 119.3978 | 2.6083 | 1.5122 | 1.0961 | 83.0000 | 3155.0000 |
| 11 | large_inputs_nisb__w512_3cube_20261003_141341.step400000.cc_threshold | artifacts (missing)checkpoint | 0.4605 | cc_threshold(logit=+9, thr=0.8581) | 118.1630 | 2.5887 | 1.5511 | 1.0376 | 122.0000 | 4213.0000 |
| 12 | large_inputs_nisb__w256_1cube_20261003_141051.step400000.cc_threshold | artifacts (missing)checkpoint | 0.4511 | cc_threshold(logit=+6, thr=0.7685) | 115.7432 | 2.5333 | 1.5011 | 1.0321 | 117.0000 | 2982.0000 |
| 13 | ds__1cube_subpixel_20260822_001229.step100000.cc_threshold | artifactscheckpoint | 0.4421 | cc_threshold(logit=+6, thr=0.7685) | 113.4410 | 2.8330 | 1.6373 | 1.1957 | 92.0000 | 4443.0000 |
| 14 | large_inputs_nisb__w256_5cube_20261003_141051.step400000.cc_threshold | artifacts (missing)checkpoint | 0.4420 | cc_threshold(logit=+6, thr=0.7685) | 113.4045 | 2.4024 | 1.4770 | 0.9254 | 166.0000 | 2764.0000 |
| 15 | large_inputs_nisb__w512_3cube_20261003_141341.step500000.cc_threshold | artifacts (missing)checkpoint | 0.4353 | cc_threshold(logit=+9, thr=0.8581) | 111.6946 | 2.5930 | 1.5455 | 1.0475 | 170.0000 | 4453.0000 |
| 16 | large_inputs_nisb__w256_1cube_20261003_141051.step300000.cc_threshold | artifacts (missing)checkpoint | 0.4293 | cc_threshold(logit=+7, thr=0.8022) | 110.1561 | 2.9041 | 1.6224 | 1.2817 | 102.0000 | 3553.0000 |
| 17 | large_inputs_nisb__w256_5cube_20261003_141051.step300000.cc_threshold | artifacts (missing)checkpoint | 0.4092 | cc_threshold(logit=+6, thr=0.7685) | 104.9956 | 2.4417 | 1.5206 | 0.9210 | 219.0000 | 3080.0000 |
| 18 | large_inputs_nisb__w256_3cube_20261003_141055.step500000.cc_threshold | artifacts (missing)checkpoint | 0.4003 | cc_threshold(logit=+6, thr=0.7685) | 102.7066 | 2.5133 | 1.5375 | 0.9758 | 209.0000 | 3413.0000 |
| 19 | large_inputs_nisb__w512_1cube_20261003_141329.step100000.cc_threshold | artifacts (missing)checkpoint | 0.3946 | cc_threshold(logit=+8, thr=0.8320) | 101.2380 | 2.7283 | 1.6320 | 1.0963 | 186.0000 | 5963.0000 |
| 20 | large_inputs_nisb__w256_1cube_20261003_141051.step200000.cc_threshold | artifacts (missing)checkpoint | 0.3730 | cc_threshold(logit=+6, thr=0.7685) | 95.6957 | 2.9294 | 1.6321 | 1.2973 | 170.0000 | 3147.0000 |
| 21 | large_inputs_nisb__w256_3cube_20261003_141055.step400000.cc_threshold | artifacts (missing)checkpoint | 0.3720 | cc_threshold(logit=+7, thr=0.8022) | 95.4444 | 3.3702 | 1.8155 | 1.5547 | 97.0000 | 5038.0000 |
| 22 | large_inputs_nisb__w256_5cube_20261003_141051.step200000.cc_threshold | artifacts (missing)checkpoint | 0.3673 | cc_threshold(logit=+6, thr=0.7685) | 94.2405 | 3.0949 | 1.7117 | 1.3833 | 161.0000 | 3594.0000 |
| 23 | large_inputs_nisb__w256_3cube_20261003_141055.step300000.cc_threshold | artifacts (missing)checkpoint | 0.3005 | cc_threshold(logit=+7, thr=0.8022) | 77.1089 | 3.7171 | 1.9754 | 1.7417 | 158.0000 | 5788.0000 |
| 24 | large_inputs_nisb__w256_1cube_20261003_141051.step100000.cc_threshold | artifacts (missing)checkpoint | 0.2910 | cc_threshold(logit=+7, thr=0.8022) | 74.6748 | 3.9092 | 1.9708 | 1.9383 | 135.0000 | 5186.0000 |
| 25 | large_inputs_nisb__w512_1cube_20261003_141329.step200000.cc_threshold | artifacts (missing)checkpoint | 0.2857 | cc_threshold(logit=+9, thr=0.8581) | 73.3052 | 3.0646 | 1.7940 | 1.2706 | 366.0000 | 8088.0000 |
| 26 | large_inputs_nisb__w256_5cube_20261003_141051.step100000.cc_threshold | artifacts (missing)checkpoint | 0.2816 | cc_threshold(logit=+7, thr=0.8022) | 72.2485 | 4.0219 | 2.0639 | 1.9580 | 132.0000 | 5649.0000 |
| 27 | large_inputs_nisb__w512_1cube_20261003_141329.step300000.cc_threshold | artifacts (missing)checkpoint | 0.2635 | cc_threshold(logit=+10, thr=0.8808) | 67.6067 | 4.0009 | 2.0469 | 1.9540 | 222.0000 | 11209.0000 |
| 28 | large_inputs_nisb__w512_1cube_20261003_141329.step500000.cc_threshold | artifacts (missing)checkpoint | 0.2521 | cc_threshold(logit=+11, thr=0.9002) | 64.6771 | 4.1590 | 2.3413 | 1.8177 | 228.0000 | 19999.0000 |
| 29 | large_inputs_nisb__w512_1cube_20261003_141329.step400000.cc_threshold | artifacts (missing)checkpoint | 0.2440 | cc_threshold(logit=+10, thr=0.8808) | 62.6102 | 3.1404 | 1.8015 | 1.3389 | 464.0000 | 8427.0000 |
| 30 | large_inputs_nisb__w256_3cube_20261003_141055.step200000.cc_threshold | artifacts (missing)checkpoint | 0.2400 | cc_threshold(logit=+6, thr=0.7685) | 61.5806 | 3.8621 | 2.1061 | 1.7560 | 252.0000 | 6565.0000 |
| 31 | large_inputs_nisb__w256_3cube_20261003_141055.step100000.cc_threshold | artifacts (missing)checkpoint | 0.0584 | cc_threshold(logit=+6, thr=0.7685) | 14.9839 | 6.6837 | 3.0897 | 3.5940 | 272.0000 | 12181.0000 |
| 32 | ds__1cube_subpixel_20260822_001229.step100000.mws_blockwise | artifactscheckpoint | 0.0099 | mws_blockwise(repulsive_stride=1, block=[512, 512, 256]) | 2.5368 | 2.1912 | 2.0882 | 0.1030 | 4630.0000 | 35886.0000 |
| 33 | large_inputs_nisb__w256_1cube_20261003_141051.step100000.mws_blockwise | artifacts (missing)checkpoint | 0.0069 | mws_blockwise(repulsive_stride=1, block=[512, 512, 256]) | 1.7600 | 2.2147 | 2.0591 | 0.1556 | 5541.0000 | 37546.0000 |
| 34 | large_inputs_nisb__w512_1cube_20261003_141329.step100000.mws_blockwise | artifacts (missing)checkpoint | 0.0063 | mws_blockwise(repulsive_stride=1, block=[512, 512, 256]) | 1.6143 | 1.9880 | 1.8450 | 0.1430 | 4316.0000 | 29873.0000 |
| 35 | large_inputs_nisb__w256_1cube_20261003_141051.step100000.mws_blockwise_1024 | artifacts (missing)checkpoint | 0.0035 | mws_blockwise(repulsive_stride=1, block=[1024, 1024, 1350]) | 0.9017 | 1.7511 | 1.6130 | 0.1380 | 5182.0000 | 33771.0000 |
| 36 | large_inputs_nisb__w512_1cube_20261003_141329.step100000.mws_blockwise_1024 | artifacts (missing)checkpoint | 0.0028 | mws_blockwise(repulsive_stride=1, block=[1024, 1024, 1350]) | 0.7073 | 1.5409 | 1.4025 | 0.1385 | 4046.0000 | 26030.0000 |
- ↑ nerl
- Normalised expected run length: ERL along the traced skeletons divided by the maximum possible.
- ↑ erl_um
- Expected run length in micrometres.
- ↓ voi_sum
- voi_split + voi_merge.
- ↓ voi_split
- Variation of information on the skeleton nodes, split term H(prediction | truth): over-segmentation.
- ↓ voi_merge
- Variation of information on the skeleton nodes, merge term H(truth | prediction): under-segmentation.
- ↓ n_non0_mergers
- Number of (merging segment, skeleton) pairs where one segment spans several skeletons.
- ↓ n_splits
- Number of skeleton edges the segmentation cuts.
zebrafinch_neurite_tracing
The artifact/checkpoint/view links below point to locations on the Janelia cluster and will only work on the Janelia network.
Views: neuroglancer views for every row below
Region:
- zebrafinch_benchmark_test 5300x9300x9700
| # | model | links | skeleton_erl.nerl (higher is better) | postprocess | skeleton_erl.erl_um | skeleton_erl.voi_sum | skeleton_erl.voi_split | skeleton_erl.voi_merge | skeleton_erl.n_non0_mergers | skeleton_erl.n_splits |
|---|---|---|---|---|---|---|---|---|---|---|
| 1 | ffn_januszewski2018.identity | artifacts | 0.0696 | Identity | 16.7471 | 2.2558 | 1.0679 | 1.1879 | 3064.0000 | 6472.0000 |
- ↑ nerl
- Normalised expected run length: ERL along the traced skeletons divided by the maximum possible.
- ↑ erl_um
- Expected run length in micrometres.
- ↓ voi_sum
- voi_split + voi_merge.
- ↓ voi_split
- Variation of information on the skeleton nodes, split term H(prediction | truth): over-segmentation.
- ↓ voi_merge
- Variation of information on the skeleton nodes, merge term H(truth | prediction): under-segmentation.
- ↓ n_non0_mergers
- Number of (merging segment, skeleton) pairs where one segment spans several skeletons.
- ↓ n_splits
- Number of skeleton edges the segmentation cuts.
zebrafinch_neurite_tracing_11um
The artifact/checkpoint/view links below point to locations on the Janelia cluster and will only work on the Janelia network.
Views: neuroglancer views for every row below
Region:
- zebrafinch_11um_test 540x1200x1200
| # | model | links | skeleton_erl.nerl (higher is better) | postprocess | skeleton_erl.erl_um | skeleton_erl.voi_sum | skeleton_erl.voi_split | skeleton_erl.voi_merge | skeleton_erl.n_non0_mergers | skeleton_erl.n_splits |
|---|---|---|---|---|---|---|---|---|---|---|
| 1 | lmd2__2f_affinity_unetr_simmim_lvd_20261020_200659.step100000.identity_8nm | artifactscheckpoint | 0.7860 | Identity | 14.0808 | 0.2533 | 0.1510 | 0.1023 | 21.0000 | 62.0000 |
| 2 | lmd2__2g_affinity_unetr_simmim_scratch_20261020_124554.step100000.identity_8nm | artifactscheckpoint | 0.7791 | Identity | 13.9573 | 0.2641 | 0.1618 | 0.1023 | 21.0000 | 55.0000 |
| 3 | lmd2__2a_affinity_unetr_lvd_20261018_215936.step100000.identity_8nm | artifactscheckpoint | 0.7784 | Identity | 13.9440 | 0.2738 | 0.1715 | 0.1024 | 21.0000 | 55.0000 |
| 4 | lmd2__2h_affinity_unetr_simmim_lvd_mask60_g2_20261021_071125.step100000.identity_8nm | artifactscheckpoint | 0.7742 | Identity | 13.8692 | 0.2909 | 0.1887 | 0.1022 | 21.0000 | 70.0000 |
| 5 | ffn_januszewski2018.identity | artifacts | 0.7585 | Identity | 13.5876 | 0.4019 | 0.2237 | 0.1783 | 21.0000 | 18.0000 |
| 6 | lmd2__2b_affinity_unetr_scratch_20261018_215937.step100000.identity_8nm | artifactscheckpoint | 0.7432 | Identity | 13.3135 | 0.3354 | 0.2330 | 0.1024 | 21.0000 | 62.0000 |
| 7 | lmd2__2f_affinity_unetr_simmim_lvd_20261020_200659.step100000.mws_blockwise | artifactscheckpoint | 0.6030 | mws_blockwise(repulsive_stride=1, block=[1024, 2048, 2048]) | 10.8016 | 0.8117 | 0.7385 | 0.0732 | 19.0000 | 176.0000 |
| 8 | lmd2__2a_affinity_unetr_lvd_20261018_215936.step100000.mws_blockwise | artifactscheckpoint | 0.5508 | mws_blockwise(repulsive_stride=1, block=[1024, 2048, 2048]) | 9.8673 | 1.0890 | 1.0310 | 0.0580 | 19.0000 | 243.0000 |
| 9 | lmd2__2h_affinity_unetr_simmim_lvd_mask60_g2_20261021_071125.step100000.mws_blockwise | artifactscheckpoint | 0.5431 | mws_blockwise(repulsive_stride=1, block=[1024, 2048, 2048]) | 9.7286 | 1.0754 | 1.0090 | 0.0664 | 19.0000 | 242.0000 |
| 10 | lmd2__2e_mtlsd_unetr_lvd_w10_20261019_004951.step100000.identity_8nm | artifactscheckpoint | 0.4721 | Identity | 8.4566 | 1.2044 | 1.1361 | 0.0683 | 23.0000 | 232.0000 |
| 11 | lmd2__2g_affinity_unetr_simmim_scratch_20261020_124554.step100000.mws_blockwise | artifactscheckpoint | 0.4621 | mws_blockwise(repulsive_stride=1, block=[1024, 2048, 2048]) | 8.2790 | 1.4902 | 1.4349 | 0.0554 | 21.0000 | 360.0000 |
| 12 | lmd2__2b_affinity_unetr_scratch_20261018_215937.step100000.mws_blockwise | artifactscheckpoint | 0.4150 | mws_blockwise(repulsive_stride=1, block=[1024, 2048, 2048]) | 7.4347 | 1.8591 | 1.8068 | 0.0523 | 19.0000 | 462.0000 |
| 13 | lmd2__2e_mtlsd_unetr_lvd_w10_20261019_004951.step100000.mws_blockwise | artifactscheckpoint | 0.2368 | mws_blockwise(repulsive_stride=1, block=[1024, 2048, 2048]) | 4.2417 | 2.6349 | 2.6006 | 0.0342 | 16.0000 | 791.0000 |
| 14 | lmd2__2d_sam_musam128_lvd_20261018_231122.step100000.identity | artifactscheckpoint | 0.1715 | Identity | 3.0729 | 3.5992 | 0.9734 | 2.6258 | 45.0000 | 9.0000 |
| 15 | lmd2__2c_mtlsd_unetr_lvd_20261018_215102.step100000.identity_8nm | artifactscheckpoint | 0.1465 | Identity | 2.6239 | 3.4507 | 3.4123 | 0.0384 | 17.0000 | 987.0000 |
| 16 | lmd2__2d_sam_musam128_lvd_20261018_231122.step100000.identity_8nm | artifactscheckpoint | 0.0587 | Identity | 1.0508 | 3.9127 | 0.5822 | 3.3305 | 52.0000 | 9.0000 |
| 17 | lmd2__2c_mtlsd_unetr_lvd_20261018_215102.step100000.mws_blockwise | artifactscheckpoint | 0.0347 | mws_blockwise(repulsive_stride=1, block=[1024, 2048, 2048]) | 0.6222 | 5.0978 | 5.0919 | 0.0060 | 14.0000 | 1984.0000 |
- ↑ nerl
- Normalised expected run length: ERL along the traced skeletons divided by the maximum possible.
- ↑ erl_um
- Expected run length in micrometres.
- ↓ voi_sum
- voi_split + voi_merge.
- ↓ voi_split
- Variation of information on the skeleton nodes, split term H(prediction | truth): over-segmentation.
- ↓ voi_merge
- Variation of information on the skeleton nodes, merge term H(truth | prediction): under-segmentation.
- ↓ n_non0_mergers
- Number of (merging segment, skeleton) pairs where one segment spans several skeletons.
- ↓ n_splits
- Number of skeleton edges the segmentation cuts.
Generated from the records under leaderboard/ by mia-evals leaderboard, from the same tables as each task's README.md.