Model zoo¶
spaCR ships a catalogue of trained models and fetches them on demand. Name a
key in a settings file — pathogen_model: toxoplasma_pv_v1 — and the model
is downloaded and checksum-verified the first time it is needed, or open
Model Zoo from the Make Masks masthead to browse and install them.
Every published entry carries a SHA-256. An entry without one is refused rather than installed, because a truncated or substituted checkpoint cannot be told from the real one.
Model |
Training data |
Hold-out performance |
|---|---|---|
|
anti-Toxoplasma-biotin and DsRed PV lumen; 229 images from 2 datasets, 104 round-1 and 125 newly curated |
F1 0.793 against stock cpsam_v2’s 0.402 on the 486 strict hold-out fields at IoU 0.5; AJI 0.558 against 0.052 |
|
crystal violet plaque wells; 184 wells from 3 datasets, 95 in-house and 89 literature |
F1 0.707 against stock cpsam’s 0.258 on the 498 strict hold-out fields at IoU 0.5; AJI 0.657 against 0.265 |
|
488 curated fields across four domains – 298 wells cropped from published figures, 96 phone-camera wells, 67 PFA and 27 methanol-fixed whole-well microscope scans; 27,582 plaques |
F1 0.455 against stock cpsam_v2’s 0.189 on the 282 strict hold-out fields at IoU 0.5; AJI 0.514 against 0.181 |
|
whole-plate and multi-well crystal violet images; 562 images from 1 dataset, 190 of them with no well in them |
mAP50 0.993 on its own held-out split; on the test set shared with v2 it scores mAP50 0.8838, against v2’s 0.9457 |
|
plate images and literature figures; 1,070 train / 254 val / 129 test, split by PMC article so no paper is in two sets; training data at einarolafsson/toxoplasma-plaque-well-detector-dataset |
mAP50 0.9457 and mAP50-95 0.8341 against v1’s (yolo_welldetect_v3.pt) 0.8838 and 0.7630 on the SAME test set; stock YOLO has no plaque-well class, so v1 is the baseline |
|
Toxoplasma PV masks predicted from the HOST CELL MASK channel alone; 2567 training and 463 held-out fields, split by well, hosts HFF/HeLa/THP1 |
F1 0.606 against 0.021 for stock cpsam_v2 on 463 well-grouped held-out fields, at IoU 0.5 |
|
anti-Toxoplasma-biotin and DsRed PV lumen; 556 curated images accumulated over five rounds |
F1 0.713 against stock cpsam_v2’s 0.364 on the 258 strict hold-out fields at IoU 0.5; AJI 0.451 against 0.049 |
|
the 556 curated PV fields of round 5, split 437 train / 108 validation / 11 test; training data at einarolafsson/toxoplasma-pv-segmentation-dataset |
F1 0.750 against stock cpsam_v2’s 0.390 on the 269 strict hold-out fields at IoU 0.5; AJI 0.528 against 0.055 |
|
round 6’s 556 curated fields plus 80 hand-curated fields of a new plate (Anu revision, Replication09182026 plate 1); 502 train / 123 validation / 11 test; training data at einarolafsson/toxoplasma-pv-segmentation-dataset-r7 |
F1 0.761 against stock cpsam_v2’s 0.377 on the 209 strict hold-out fields at IoU 0.5; AJI 0.582 against 0.040 |
|
285 curated PV fields (lab fields and published-figure crops), 13,814 vacuoles; 202 train / 33 validation / 50 test grouped by acquisition and paper |
F1 0.733 against stock cpsam_v2’s 0.383 on the 481 strict hold-out fields at IoU 0.5; AJI 0.585 against 0.051 |
|
11,007 transmitted-light fields from 14 public datasets, split by acquisition 6,778 train / 2,030 validation / 2,199 test; training data at einarolafsson/live-cell-segmentation-dataset |
on the datasets stock cpsam_v2 never trained on, F1 0.960 against 0.885 at IoU 0.5; over all 2,199 test fields, 0.694 against 0.738, because stock trained on LIVECell and YeaZ and wins on LIVECell |
|
nuclei predicted from the HOST CELL MASK channel alone; 453 well-grouped held-out fields, hosts HFF/HeLa/THP1 |
F1 0.888 against 0.201 for stock cpsam_v2 on 453 well-grouped held-out fields, at IoU 0.5 |
|
the HOST CELL outline predicted from the Hoechst (nuclear) channel alone; 2,578 training fields and 451 held-out test fields, split by well so no well is on both sides |
F1 0.870 against stock cpsam_v2’s 0.301 on 451 held-out fields at IoU 0.5 – a delta of 0.569 |
|
Toxoplasma PV masks predicted from the HOECHST channel alone; 2567 training and 463 held-out fields, split by well, hosts HFF/HeLa/THP1 |
F1 0.569 against 0.002 for stock cpsam_v2 on 463 well-grouped held-out fields, at IoU 0.5 |
|
496 curated plaque fields, including 34 reviewed empty negatives and 71 Gel Doc wells; 100 epochs; fixed physical-plate/source groups |
F1 0.732 against stock cpsam’s 0.128 on the 232 strict hold-out fields at IoU 0.5; AJI 0.628 against 0.173 |
|
663 curated plaque fields from four domains (published-figure wells, Gel Doc wells, lab plates, staged literature crops), 32,717 plaques; 335 train / 65 validation / 263 test |
F1 0.765 against stock cpsam_v2’s 0.338 on the 329 strict hold-out fields at IoU 0.5; AJI 0.662 against 0.261 |
|
1,507 curated plaque fields of 2026-10-15 (Gel Doc wells, lab plates, published figures, earlier curation rounds); 1,201 train / 153 validation / 153 test |
F1 0.816 against stock cpsam_v2’s 0.227 on the 153 strict hold-out fields at IoU 0.5; AJI 0.737 against 0.178 |
|
452 reviewed training images; 124 validation images; physical plate and figure groups; 150 epochs; YOLO11n v3 initialization |
Stock was not evaluated in this run; see the named incumbent comparison on the model card |
Each model is hosted in a repository under its author’s own Hugging Face
account; contributing a model does not require write access to another
account’s repository.
spacr.model_zoo’s publish_model performs the upload and prints the
catalogue row to add.
Choose a model from a form¶
A model setting in Mask generation, Make Masks and Plaque Assay has a
Model zoo… button. Its list is grouped by origin under five headings —
the stock Cellpose-SAM weights, spaCR’s own models, unvetted spaCR community
uploads, Cellpose models published on bioimage.io, and the Cellpose 3
backend’s cyto, cyto2, cyto3 and nuclei models. Click a heading to show or
fold its rows. Download fetches the selected model, or Install its
backend; Use this model then writes the model into the setting. A model that needs another backend, such as
Cellpose 3 or Cellpose-DINO, says so on its row and offers to install that
backend into its own environment under ~/.spacr/backends; spaCR’s own
environment is not changed. A row spaCR cannot run says so instead of being
offered.
Opened from Mask generation, the popup also has Measure diameters…: it
estimates the cell, nucleus and pathogen diameters from a few fields of your
own images and can write them into the settings. See
spacr.qt.prerun.DiameterDialog.
Per-model detail¶
Toxoplasma PV v1¶
Architecture. Cellpose-SAM (cpsam_v2)
Trained on. Toxoplasma tachyzoite parasitophorous vacuoles stained with goat anti-Toxoplasma-biotin, and tachyzoites expressing DsRed in the PV lumen. Round 2: 229 training images (round 1’s 104 plus 125 newly curated RH and ME49 fields), 100 epochs, base cpsam_v2
Measured. F1 0.793 against stock cpsam_v2’s 0.402 on the 486 strict hold-out fields at IoU 0.5; AJI 0.558 against 0.052
Metric |
This model |
Stock |
Difference |
|---|---|---|---|
F1 |
0.7930 |
0.4016 |
0.3914 |
Boundary F1 (2 px) |
0.8271 |
0.2584 |
0.5687 |
Precision |
0.8604 |
0.2921 |
0.5683 |
Recall |
0.7354 |
0.6427 |
0.0927 |
AJI |
0.5579 |
0.0517 |
0.5062 |
Dice |
0.7941 |
0.1149 |
0.6792 |
Evaluation set: gt_rescore_20261011; version strict hold-out, pooled; 486 scored fields and 14315 annotated objects.
F1 0.864 at IoU 0.5 against 0.713 for stock cpsam on the 11 wells round 1 also held out (round 1 scored 0.867); AJI 0.809 against 0.426
accuracy falls sharply above IoU 0.8 – suited to counting and area rather than precise morphometry
the held-out literature scorecard is pending a stock-seeded re-curation; on the current literature set, whose truth leans toward this model’s lineage, it ties stock Cellpose-SAM on detection (F1 0.403 against 0.400)
Re-scored 2026-10-11 on curated ground truth (spaCR pipeline call, defaults; receipt features/data/model_rescore_curated_gt_2026-10-11/receipt.json): strict hold-out, fields this model never saw, F1 0.793 at IoU 0.5 on n = 486 of 655 fields (n differs between models); common subset unseen by every pv model, n = 163, F1 0.783 (95% bootstrap CI 0.710 to 0.830). Without the 98 paged round-3 fields (n = 65): F1 0.732. Earlier numbers in this row are superseded by this re-score where they differ. Default unchanged.
Published as einarolafsson/toxoplasma-pv-segmentation-cpsam, as cpsam_v2_toxo_r2.
SHA-256 182d8cf6b32c7b9ef2917c85870d188486e5e119f05e9c5c1f07652f6859f2d0.
Toxoplasma Plaque v1¶
Architecture. Cellpose-SAM (cpsam)
Trained on. Toxoplasma gondii plaque assays; round 3, evaluated in-domain (NAS) and against a literature generalisation set
Measured. F1 0.707 against stock cpsam’s 0.258 on the 498 strict hold-out fields at IoU 0.5; AJI 0.657 against 0.265
Metric |
This model |
Stock |
Difference |
|---|---|---|---|
F1 |
0.7074 |
0.2581 |
0.4493 |
Boundary F1 (2 px) |
0.6794 |
0.3269 |
0.3525 |
Precision |
0.8298 |
0.2050 |
0.6248 |
Recall |
0.6165 |
0.3484 |
0.2680 |
AJI |
0.6572 |
0.2649 |
0.3923 |
Dice |
0.8191 |
0.4667 |
0.3524 |
Evaluation set: gt_rescore_20261011; version strict hold-out, pooled; 498 scored fields and 19993 annotated objects.
F1 0.856 in-domain and 0.806 on the literature set (3-fold cross-validated, SD 0.020), against 0.718 for round 1
round 3 trades precision (0.939 down to 0.858) for recall (0.631 up to 0.811) on the literature set, which is the right direction for a counting assay
PREFER THIS ONE FOR MICROSCOPE-ONLY WORK. On the round-5 test split it scores 0.836 on PFA-fixed wells against round 5’s 0.808, at precision 0.93 against 0.81. For mixed sources, or any phone-camera image, use toxoplasma_plaque_v2, which round 3 cannot handle at all (0.249 there)
Re-scored 2026-10-11 on curated ground truth (spaCR pipeline call, defaults; receipt features/data/model_rescore_curated_gt_2026-10-11/receipt.json): strict hold-out, fields this model never saw, F1 0.707 at IoU 0.5 on n = 498 of 562 fields (n differs between models); common subset unseen by every plaque model, n = 114, F1 0.402 (95% bootstrap CI 0.300 to 0.504). Without the v3-seeded Gel Doc truth (n = 81): F1 0.472. Earlier numbers in this row are superseded by this re-score where they differ. 263-field set seen counts: v1 54, v2 178, v3 137, v4 0; GRA8 Plate1 (hold-out of v3) is the same physical plate as ReimagedPlate1 (training data of v4). Default unchanged.
Published as einarolafsson/toxoplasma-plaque-segmentation-cpsam, as cpsam_plaque_r3.
SHA-256 eeecd2d6cd5cbb4dddee71564d5f460d26bb07ac125e0b494b7502fea4292d5d.
Toxoplasma Plaque v2 (round 5)¶
Architecture. Cellpose-SAM (cpsam_v2)
Trained on. Toxoplasma plaque assays stained with crystal violet, from three microscopes and from published figures. Round 5: 332 training fields grouped by figure and by plate so none straddles the split, 100 epochs, base cpsam_v2, empty wells kept as negatives
Measured. F1 0.455 against stock cpsam_v2’s 0.189 on the 282 strict hold-out fields at IoU 0.5; AJI 0.514 against 0.181
Metric |
This model |
Stock |
Difference |
|---|---|---|---|
F1 |
0.4548 |
0.1891 |
0.2657 |
Boundary F1 (2 px) |
0.5679 |
0.2447 |
0.3232 |
Precision |
0.5004 |
0.2007 |
0.2997 |
Recall |
0.4169 |
0.1787 |
0.2381 |
AJI |
0.5145 |
0.1812 |
0.3332 |
Dice |
0.6961 |
0.3141 |
0.3820 |
Evaluation set: gt_rescore_20261011; version strict hold-out, pooled; 282 scored fields and 8130 annotated objects.
no longer the default: toxoplasma_plaque_v3 replaced it on 2026-10-10 (F1 0.844 against 0.799 on 263 held-out fields; this model scores 0.013 on Gel Doc wells), and toxoplasma_plaque_v4 on 2026-10-11; kept so recorded runs reproduce
COMPLEMENTS toxoplasma_plaque_v1 rather than replacing it: prefer v1 (round 3) for microscope-only work, where it scores 0.836 against this model’s 0.808 on PFA-fixed wells and is far more precise; prefer this one for mixed or unknown sources
the only plaque model trained on phone-camera wells – F1 0.415 against round 3’s 0.249, though recall there is 0.296, so it still misses most plaques on phone images and is not yet a counting tool
it did NOT clear the promotion bar of 0.02 literature F1 fixed before the run (it came in at -0.001), so round 3 remains production
balanced precision/recall (0.81/0.83) where round 3 is lopsided (0.93/0.73): round 3’s low recall systematically UNDERCOUNTS, which matters more than F1 for a counting assay
hallucinates 2 objects across 6 blank-lawn wells where round 3 hallucinates 19
first plaque model on cpsam_v2; rounds 1-4 used cpsam v1, so base and data changed together and the gap to round 3 is not attributable to the extra curation alone
training data: https://huggingface.co/datasets/einarolafsson/toxoplasma-plaque-dataset
Re-scored 2026-10-11 on curated ground truth (spaCR pipeline call, defaults; receipt features/data/model_rescore_curated_gt_2026-10-11/receipt.json): strict hold-out, fields this model never saw, F1 0.455 at IoU 0.5 on n = 282 of 562 fields (n differs between models); common subset unseen by every plaque model, n = 114, F1 0.360 (95% bootstrap CI 0.249 to 0.472). Without the v3-seeded Gel Doc truth (n = 81): F1 0.406. Earlier numbers in this row are superseded by this re-score where they differ. 263-field set seen counts: v1 54, v2 178, v3 137, v4 0; GRA8 Plate1 (hold-out of v3) is the same physical plate as ReimagedPlate1 (training data of v4). Default unchanged.
Published as einarolafsson/toxoplasma-plaque-segmentation-cpsam-r5, as cpsam_plaque_r5.
SHA-256 0927023a745ac6a19bae0ec72c89b7b864a4ff8d047a41f3f1e9767e1a4d0600.
Toxoplasma Plaque Well Detector v1¶
Architecture. YOLO11n
Trained on. whole-plate and multi-well Toxoplasma plaque-assay images; yolo11n base, 150 epochs, batch 16, imgsz 640
Measured. mAP50 0.993 on its own held-out split; on the test set shared with v2 it scores mAP50 0.8838, against v2’s 0.9457
the 0.993 is measured on v3’s OWN split, which is easier than the set v2 is measured on; on that shared set this model scores mAP50 0.8838 against v2’s 0.9457, so v2 is the better detector
kept because the published plaque corpus was measured with these weights, so results in the paper trace back to this row
locates WELLS, not plaques; it is the front half of a two-stage pipeline with toxoplasma_plaque_v1, and the well it finds also gives the diameter that makes areas comparable across microscopes
Published as einarolafsson/toxoplasma-plaque-well-detector-yolo11, as yolo_welldetect_v3.pt.
SHA-256 b826058754fb5d4df36c3a7283aac049015cbb044b5ef096c55d19f37172a50c.
Toxoplasma Plaque Well Detector v2¶
Architecture. YOLO26n (ultralytics 8.4.155)
Trained on. whole-plate and multi-well Toxoplasma plaque-assay images plus 939 newly reviewed PMC figures, accepted boxes and confirmed negatives alike; yolo26n base, best validation mAP50-95 at epoch 28
Measured. mAP50 0.9457 and mAP50-95 0.8341 against v1’s (yolo_welldetect_v3.pt) 0.8838 and 0.7630 on the SAME test set; stock YOLO has no plaque-well class, so v1 is the baseline
on the shared test set it beats v1 (the v3 weights) on every measure, and cuts false boxes on no-well figures from 152 to 49
84 of the 129 test images contain no well at all, which is what the false-box count is measured on
locates WELLS, not plaques; the front half of a two-stage pipeline with the plaque segmentation model
the repository publishes this weight as weights/best.pt; spaCR saves it under the name above so two detectors cannot both land as best.pt
Published as einarolafsson/toxoplasma-plaque-well-detector-yolo26, as yolo_welldetect_v4.pt.
SHA-256 f2a1e1110f09b2a1d5ef5545adaba7c57f1158669d0bfc50d8fabe9f86da30c7.
Toxoplasma from Cell Mask (cross-channel)¶
Architecture. Cellpose-SAM (cpsam_v2)
Trained on. cross-channel: given the host cell image, predicts where the Toxoplasma parasitophorous vacuoles are, with no parasite stain. 100 epochs, base cpsam_v2, AdamW lr 1e-5, targets are PV-regenerated masks
Measured. F1 0.606 against 0.021 for stock cpsam_v2 on 463 well-grouped held-out fields, at IoU 0.5
F1 0.606, AJI 0.494, Dice 0.610 at IoU 0.5 against stock cpsam_v2’s 0.021/0.008/0.020 – stock cannot do this task at all
per host: HeLa 0.711, HFF 0.557, THP1 0.465; THP1 is the weak case
the held-out split selects the checkpoint, so it is validation data rather than an independent test set
accuracy falls above IoU 0.8 – suited to counting, occupancy and area rather than precise morphometry
Published as einarolafsson/toxoplasma-from-cellmask-cpsam, as toxoplasma_from_cellmask_pv.
SHA-256 481dfccc1a68cc594aafcb71088efc25b5f5c6a6240e52902c0089759b3149ab.
Toxoplasma PV v2 (round 5)¶
Architecture. Cellpose-SAM (cpsam_v2)
Trained on. Toxoplasma tachyzoite parasitophorous vacuoles stained with goat anti-Toxoplasma-biotin, and tachyzoites expressing DsRed in the PV lumen (RH and ME49). Round 5: 556 images, 100 epochs, base cpsam_v2
Measured. F1 0.713 against stock cpsam_v2’s 0.364 on the 258 strict hold-out fields at IoU 0.5; AJI 0.451 against 0.049
Metric |
This model |
Stock |
Difference |
|---|---|---|---|
F1 |
0.7130 |
0.3638 |
0.3492 |
Boundary F1 (2 px) |
0.7340 |
0.2301 |
0.5040 |
Precision |
0.7658 |
0.2560 |
0.5098 |
Recall |
0.6670 |
0.6287 |
0.0383 |
AJI |
0.4507 |
0.0488 |
0.4019 |
Dice |
0.6782 |
0.1116 |
0.5666 |
Evaluation set: gt_rescore_20261011; version strict hold-out, pooled; 258 scored fields and 5766 annotated objects.
supersedes toxoplasma_pv_v1 (round 2, 229 images): more than twice the training data and cross-validated rather than single-split
5-fold CV over 619 pairs: F1 0.817 (SD 0.036), AJI 0.714, Dice 0.802
per-dataset variance is real – F1 ranges ~0.74 to ~0.93 by screen
accuracy falls above IoU 0.8 – suited to counting and area rather than precise morphometry
Re-scored 2026-10-11 on curated ground truth (spaCR pipeline call, defaults; receipt features/data/model_rescore_curated_gt_2026-10-11/receipt.json): strict hold-out, fields this model never saw, F1 0.713 at IoU 0.5 on n = 258 of 655 fields (n differs between models); common subset unseen by every pv model, n = 163, F1 0.761 (95% bootstrap CI 0.691 to 0.806). Without the 98 paged round-3 fields (n = 65): F1 0.742. Earlier numbers in this row are superseded by this re-score where they differ. Default unchanged.
Published as einarolafsson/toxoplasma-pv-segmentation-cpsam-r5, as cpsam_v2_toxo_r5.
SHA-256 17c689e3b117745561e20a885c2a2a998ed360fa97cac8c0446316ae5905c10f.
Toxoplasma PV v3 (round 6)¶
Architecture. Cellpose-SAM (cpsam_v2)
Trained on. Toxoplasma tachyzoite parasitophorous vacuoles stained with goat anti-Toxoplasma-biotin, and tachyzoites expressing DsRed in the PV lumen (RH and ME49). Round 6 retrains round 5’s data with cellpose 4.2.1.1, 100 epochs, base cpsam_v2
Measured. F1 0.750 against stock cpsam_v2’s 0.390 on the 269 strict hold-out fields at IoU 0.5; AJI 0.528 against 0.055
Metric |
This model |
Stock |
Difference |
|---|---|---|---|
F1 |
0.7504 |
0.3899 |
0.3606 |
Boundary F1 (2 px) |
0.7826 |
0.2537 |
0.5289 |
Precision |
0.7998 |
0.2794 |
0.5204 |
Recall |
0.7068 |
0.6447 |
0.0621 |
AJI |
0.5278 |
0.0553 |
0.4725 |
Dice |
0.7499 |
0.1236 |
0.6263 |
Evaluation set: gt_rescore_20261011; version strict hold-out, pooled; 269 scored fields and 6443 annotated objects.
NEWEST IS NOT BEST HERE: round 6 does not beat round 2 on the anchor wells – 0.8602 against 0.8648 – and the PV project still promotes round 5
the shipped weights are the FINAL epoch 100, not a validation-chosen checkpoint: the training records show no checkpoint selection (recorded 2026-10-12)
5-fold cross-validation, grouped by source: F1 0.8168 +/- 0.028, AJI 0.7516, Dice 0.8424
the 11 anchor wells have been held out since round 1, so they are the only fields no PV round has ever trained on
Re-scored 2026-10-11 on curated ground truth (spaCR pipeline call, defaults; receipt features/data/model_rescore_curated_gt_2026-10-11/receipt.json): strict hold-out, fields this model never saw, F1 0.750 at IoU 0.5 on n = 269 of 655 fields (n differs between models); common subset unseen by every pv model, n = 163, F1 0.790 (95% bootstrap CI 0.726 to 0.835). Without the 98 paged round-3 fields (n = 65): F1 0.754. Earlier numbers in this row are superseded by this re-score where they differ. Default unchanged.
Published as einarolafsson/toxoplasma-pv-segmentation-cpsam-r6, as cpsam_v2_toxo_r6.
SHA-256 146ef269979b1d1ab45c11039b0ab164f68001adaa8f73f1f8f18be6fcfd060e.
Toxoplasma PV v4 (round 7)¶
Architecture. Cellpose-SAM (cpsam_v2)
Trained on. Toxoplasma tachyzoite parasitophorous vacuoles: round 6’s DsRed and anti-Toxoplasma-biotin fields plus the Toxoplasma channel of a new 384-well plate. Round 7, cellpose 4.2.1.1, 100 epochs, base cpsam_v2
Measured. F1 0.761 against stock cpsam_v2’s 0.377 on the 209 strict hold-out fields at IoU 0.5; AJI 0.582 against 0.040
Metric |
This model |
Stock |
Difference |
|---|---|---|---|
F1 |
0.7611 |
0.3771 |
0.3840 |
Boundary F1 (2 px) |
0.8053 |
0.2229 |
0.5825 |
Precision |
0.7687 |
0.2647 |
0.5040 |
Recall |
0.7538 |
0.6557 |
0.0981 |
AJI |
0.5820 |
0.0396 |
0.5424 |
Dice |
0.8348 |
0.0906 |
0.7442 |
Evaluation set: gt_rescore_20261011; version strict hold-out, pooled; 209 scored fields and 5515 annotated objects.
Scorecard metrics are means over nonempty fields, not pooled object counts; see aggregation.json.
NEWEST IS NOT BEST HERE: on 20 held-out fields of the new plate it was trained on, round 7 scores F1 0.748 against 0.899 for round 6 (recall 0.636 against 0.829) – prefer toxoplasma_pv_v3 on that plate
on the 11 anchor wells it ties round 6 (0.8536 against 0.8602); 5-fold cross-validation F1 0.8142 +/- 0.015 against round 6’s 0.8168
it traces the vacuoles it finds more tightly than round 6 (AJI 0.827 against 0.786 on the new plate) but finds fewer of them
the drop on the new plate is under investigation: round 7 also under-fits that plate’s own training fields, which points at how those fields entered training rather than at generalisation
Re-scored 2026-10-11 on curated ground truth (spaCR pipeline call, defaults; receipt features/data/model_rescore_curated_gt_2026-10-11/receipt.json): strict hold-out, fields this model never saw, F1 0.761 at IoU 0.5 on n = 209 of 655 fields (n differs between models); common subset unseen by every pv model, n = 163, F1 0.781 (95% bootstrap CI 0.725 to 0.817). Without the 98 paged round-3 fields (n = 65): F1 0.757. Earlier numbers in this row are superseded by this re-score where they differ. Default unchanged.
Published as einarolafsson/toxoplasma-pv-segmentation-cpsam-r7, as cpsam_v2_toxo_r7.
SHA-256 621a475c9bfb6865be7de12c5c4638f4509d734e0c4d39adb15ea541e48961d2.
Toxoplasma PV v5 (round 8, alternative)¶
Architecture. Cellpose-SAM (cpsam_v2, Cellpose 4.0.9)
Trained on. Toxoplasma parasitophorous vacuoles from lab immunofluorescence and DsRed fields and published-figure crops. Round 8, cellpose 4.0.9, best of 100 epochs (epoch 20), base cpsam_v2
Measured. F1 0.733 against stock cpsam_v2’s 0.383 on the 481 strict hold-out fields at IoU 0.5; AJI 0.585 against 0.051
Metric |
This model |
Stock |
Difference |
|---|---|---|---|
F1 |
0.7329 |
0.3831 |
0.3497 |
Boundary F1 (2 px) |
0.8198 |
0.2567 |
0.5631 |
Precision |
0.6459 |
0.2749 |
0.3710 |
Recall |
0.8469 |
0.6322 |
0.2146 |
AJI |
0.5846 |
0.0509 |
0.5337 |
Dice |
0.8098 |
0.1119 |
0.6979 |
Evaluation set: gt_rescore_20261011; version strict hold-out, pooled; 481 scored fields and 16213 annotated objects.
ALTERNATIVE, NOT BETTER: on the 15 test fields round 7 never saw it scores F1 0.536 against toxoplasma_pv_v4’s 0.525 – a tie; prefer toxoplasma_pv_v4
35 of the 50 test fields were in round 7’s training or validation set, so the all-field comparison (0.792 against 0.851) favours toxoplasma_pv_v4 and is not a fair test
weak on published-figure crops (F1 0.440 on 14 literature fields), as is every PV model
Re-scored 2026-10-11 on curated ground truth (spaCR pipeline call, defaults; receipt features/data/model_rescore_curated_gt_2026-10-11/receipt.json): strict hold-out, fields this model never saw, F1 0.733 at IoU 0.5 on n = 481 of 655 fields (n differs between models); common subset unseen by every pv model, n = 163, F1 0.633 (95% bootstrap CI 0.544 to 0.695). Without the 98 paged round-3 fields (n = 65): F1 0.665. Earlier numbers in this row are superseded by this re-score where they differ. LEAKED: this model trained or validated on 29 of 39 curated PV literature hold-out fields and 3 of 11 anchor wells, so its earlier hold-out scores are not hold-out scores; the re-score uses only the fields it never saw. Default unchanged.
Published as einarolafsson/toxoplasma-pv-segmentation-cpsam-r8, as cpsam_v2_toxo_r8.
SHA-256 a1d7f1a978cf4925ec60db486a00eea33e87e6fc5a6b0d0f34c6db5f132d5c1c.
Live cell v1 (phase, brightfield, DIC; alternative)¶
Architecture. Cellpose-SAM (cpsam_v2)
Trained on. unstained cells in phase contrast, brightfield and DIC: LIVECell, DeepSea, YeaZ, yeast microstructures, five Cell Tracking Challenge sets, BBBC009, BBBC030, QPI and Revvity. Base cpsam_v2, cellpose 4.2.1.1, lr 1e-5, batch 4; stopped at epoch 37 of 100
Measured. on the datasets stock cpsam_v2 never trained on, F1 0.960 against 0.885 at IoU 0.5; over all 2,199 test fields, 0.694 against 0.738, because stock trained on LIVECell and YeaZ and wins on LIVECell
an optional alternative: stock Cellpose-SAM (cpsam_v2) stays the default for label-free and live-cell segmentation
TWO STOCK COMPARISONS, NOT ONE: stock cpsam_v2 trained on LIVECell and YeaZ, so its score there is partly memorisation. On the datasets it never saw (DeepSea, the CTC sets, BBBC009, BBBC030, QPI, Revvity, yeast microstructures) this model scores F1 0.960 against 0.885
it does NOT replace stock on LIVECell-style Incucyte phase: 0.671 against 0.724 there, and it missed its own pre-registered promotion bar
by modality at IoU 0.5: brightfield 0.964 (stock 0.912), DIC +0.026 over stock, phase 0.689 (stock 0.735)
F1 0.865 on a train sample, 0.696 on validation and 0.694 on test; no per-epoch loss was recorded
Published as einarolafsson/live-cell-segmentation-cpsam, as live_cell_v1.
SHA-256 7ade69377093fe81830ddc7c52ba8618bef1fefe7d1243c01b9c1beed7fcb090.
Cross-channel nuclei-from-cellmask¶
Architecture. Cellpose-SAM (cpsam_v2)
Trained on. cross-channel: given the cell image, predicts where the nuclei are, with no nuclear stain – which frees the DAPI/Hoechst channel for another marker. 100 epochs, base cpsam_v2
Measured. F1 0.888 against 0.201 for stock cpsam_v2 on 453 well-grouped held-out fields, at IoU 0.5
F1 0.888, AJI 0.792, Dice 0.877 at IoU 0.5 against stock cpsam_v2’s 0.201/0.286/0.449
per host: HFF 0.932, HeLa 0.860, THP1 0.861
the held-out split selects the checkpoint, so it is validation data rather than an independent test set
predicts nuclei from cell morphology – expect degraded accuracy on unusual or highly confluent morphologies
Published as einarolafsson/cross-channel-nuclei-from-cellmask-cpsam, as nuclei_from_cellmask_best.
SHA-256 2675553a46e97a7bc4bd2bfe3e954954194fe71ca4e94261e752a02bf0b6eb47.
Cross-channel cell-from-hoechst¶
Architecture. Cellpose-SAM (cpsam_v2)
Trained on. Hoechst-stained nuclei paired with curated host-cell masks; fine-tuned from stock cpsam_v2, 100 epochs, best epoch 70
Measured. F1 0.870 against stock cpsam_v2’s 0.301 on 451 held-out fields at IoU 0.5 – a delta of 0.569
the counterpart of nuclei_from_cellmask_v1: that one predicts nuclei from the cell mask, this one predicts the cell from the nucleus
precision 0.944 against recall 0.806 – it misses cells rather than inventing them, which is the safer direction for counting
quote the DELTA over stock (0.569), not the ratio: stock’s mAP of 0.0575 is a near-zero denominator that makes any ratio look huge
Published as einarolafsson/cross-channel-cell-from-hoechst-cpsam, as cell_from_hoechst_best.
SHA-256 d1992433b4f2f291f73738830bb953198165fdc10719e54dae9b8c2bd430e0eb.
Toxoplasma from Hoechst (cross-channel)¶
Architecture. Cellpose-SAM (cpsam_v2)
Trained on. cross-channel: given the Hoechst/nuclear image, predicts where the Toxoplasma parasitophorous vacuoles are, with no parasite stain. 100 epochs, base cpsam_v2, AdamW lr 1e-05, targets are PV-regenerated masks
Measured. F1 0.569 against 0.002 for stock cpsam_v2 on 463 well-grouped held-out fields, at IoU 0.5
F1 0.569, AJI 0.421, Dice 0.546 at IoU 0.5 against stock cpsam_v2’s 0.002/0.006/0.016
the Hoechst route is harder than the cell-mask route – compare toxoplasma_from_cellmask_v1
the held-out split selects the checkpoint, so it is validation data rather than an independent test set
accuracy falls above IoU 0.8 – suited to counting, occupancy and area rather than precise morphometry
Published as einarolafsson/toxoplasma-from-hoechst-cpsam, as toxoplasma_from_hoechst_pv.
SHA-256 8dc05ebced3550d1a418c13d24d319e0482c742988df29a525520026cb2f0d96.
Toxoplasma Plaque v3 (Gel Doc r5, alternative)¶
Architecture. Cellpose-SAM (cpsam, Cellpose 4.0.9)
Trained on. 496 curated plaque fields, including 34 reviewed empty negatives and 71 Gel Doc wells; 100 epochs; fixed physical-plate/source groups
Measured. F1 0.732 against stock cpsam’s 0.128 on the 232 strict hold-out fields at IoU 0.5; AJI 0.628 against 0.173
Metric |
This model |
Stock |
Difference |
|---|---|---|---|
F1 |
0.7322 |
0.1280 |
0.6042 |
Boundary F1 (2 px) |
0.6211 |
0.2157 |
0.4054 |
Precision |
0.6763 |
0.0827 |
0.5936 |
Recall |
0.7981 |
0.2825 |
0.5156 |
AJI |
0.6275 |
0.1729 |
0.4547 |
Dice |
0.8105 |
0.3337 |
0.4768 |
Evaluation set: gt_rescore_20261011; version strict hold-out, pooled; 232 scored fields and 8379 annotated objects.
Recommended alternative for Gel Doc wells: on the 153 test fields of the curated 2026-10-15 set it scores object F1 0.954 on Gel Doc wells against 0.868 for the default toxoplasma_plaque_v5 (0.741 against 0.816 overall).
Alternative since 2026-10-11, when toxoplasma_plaque_v4 (base cpsam_v2) became the default by the maintainer’s choice. This model scores higher: on 263 held-out fields F1 0.844 against v4 0.815, and 0.913 against 0.804 on Gel Doc wells, through spacr.plaque.segment_plaque_image. Kept so recorded runs reproduce.
Default 2026-10-10 to 2026-10-11: F1 0.844 against toxoplasma_plaque_v2 0.799, and 0.913 against 0.013 on Gel Doc wells (features/data/model_compare_2026-10-10).
Its own release gate on independent literature crops did not pass at the time (literature F1 about 0.54 for every model).
Gel Doc score requires min_size=0; min_size=150 gives F1 0.1041.
Stock not measured; comparison baseline is incumbent r3.
Distinct from the cpsam_v2 round-5 model already registered as toxoplasma_plaque_v2.
Re-scored 2026-10-11 on curated ground truth (spaCR pipeline call, defaults; receipt features/data/model_rescore_curated_gt_2026-10-11/receipt.json): strict hold-out, fields this model never saw, F1 0.732 at IoU 0.5 on n = 232 of 562 fields (n differs between models); common subset unseen by every plaque model, n = 114, F1 0.548 (95% bootstrap CI 0.449 to 0.645). Without the v3-seeded Gel Doc truth (n = 81): F1 0.444. Earlier numbers in this row are superseded by this re-score where they differ. 263-field set seen counts: v1 54, v2 178, v3 137, v4 0; GRA8 Plate1 (hold-out of v3) is the same physical plate as ReimagedPlate1 (training data of v4). Default unchanged.
Published as einarolafsson/toxoplasma-plaque-segmentation-cpsam-r5-geldoc, as cpsam_plaque_r5_geldoc.
SHA-256 8f59850dbe09c3a252728b0a7ca38ae5b52a28f2dbaa7a068174b4b7bfac768c.
Toxoplasma Plaque v4 (round 6, cpsam_v2)¶
Architecture. Cellpose-SAM (cpsam_v2, Cellpose 4.0.9)
Trained on. Toxoplasma plaques from published-figure wells, Bio-Rad Gel Doc wells, lab plates and staged literature crops; round 6 splits, best of 190 epochs (epoch 140), base cpsam_v2
Measured. F1 0.765 against stock cpsam_v2’s 0.338 on the 329 strict hold-out fields at IoU 0.5; AJI 0.662 against 0.261
Metric |
This model |
Stock |
Difference |
|---|---|---|---|
F1 |
0.7648 |
0.3380 |
0.4267 |
Boundary F1 (2 px) |
0.6729 |
0.3539 |
0.3190 |
Precision |
0.7344 |
0.2911 |
0.4433 |
Recall |
0.7977 |
0.4030 |
0.3947 |
AJI |
0.6619 |
0.2610 |
0.4010 |
Dice |
0.8388 |
0.4291 |
0.4097 |
Evaluation set: gt_rescore_20261011; version strict hold-out, pooled; 329 scored fields and 13915 annotated objects.
No longer the default: toxoplasma_plaque_v5 replaced it on 2026-10-17 (object F1 0.816 against 0.725 on the 153 curated test fields); kept selectable so recorded runs reproduce.
The default plaque model 2026-10-11 to 2026-10-17, by the maintainer’s rule that spaCR models build on cpsam_v2. It scores below toxoplasma_plaque_v3 on the same 263 held-out fields, through spacr.plaque.segment_plaque_image: F1 0.815 against 0.844, lower on every source (Gel Doc 0.804 against 0.913); v3 stays selectable.
Fine-tuned from cpsam_v2 (2026-10-10). It replaces the first round-6 weights, fine-tuned from first-generation cpsam (sha256 00f4a86b…a0c171), which remain in the repository history.
Above toxoplasma_plaque_v2 overall (0.799) and on Gel Doc wells, where v2 scores 0.013.
Published-figure literature crops stay hard (F1 0.530).
Re-scored 2026-10-11 on curated ground truth (spaCR pipeline call, defaults; receipt features/data/model_rescore_curated_gt_2026-10-11/receipt.json): strict hold-out, fields this model never saw, F1 0.765 at IoU 0.5 on n = 329 of 562 fields (n differs between models); common subset unseen by every plaque model, n = 114, F1 0.534 (95% bootstrap CI 0.440 to 0.625). Without the v3-seeded Gel Doc truth (n = 81): F1 0.459. Earlier numbers in this row are superseded by this re-score where they differ. 263-field set seen counts: v1 54, v2 178, v3 137, v4 0; GRA8 Plate1 (hold-out of v3) is the same physical plate as ReimagedPlate1 (training data of v4). Default unchanged.
Published as einarolafsson/toxoplasma-plaque-segmentation-cpsam-r6, as cpsam_plaque_r6.
SHA-256 744aee3f71efa68f6a369a6f91fe5850dc61e35f3dc20d53dda2c04fce17b35f.
Toxoplasma Plaque v5 (round 7, cpsam_v2, default)¶
Architecture. Cellpose-SAM (cpsam_v2, Cellpose 4.2.1.1)
Trained on. Toxoplasma plaques from Gel Doc wells, lab plates, published figures and earlier curation rounds (curated dataset of 2026-10-15); 100 epochs, final epoch kept, base cpsam_v2
Measured. F1 0.816 against stock cpsam_v2’s 0.227 on the 153 strict hold-out fields at IoU 0.5; AJI 0.737 against 0.178
Metric |
This model |
Stock |
Difference |
|---|---|---|---|
F1 |
0.8157 |
0.2268 |
0.5889 |
Boundary F1 (2 px) |
0.7585 |
0.2883 |
0.4702 |
Precision |
0.8674 |
0.1894 |
0.6779 |
Recall |
0.7699 |
0.2824 |
0.4875 |
AJI |
0.7366 |
0.1784 |
0.5582 |
Dice |
0.8962 |
0.3291 |
0.5671 |
Evaluation set: plaque_curated_20261015_test; version strict hold-out, pooled; 153 scored fields and 6540 annotated objects.
The default plaque model since 2026-10-17, by the maintainer’s choice; toxoplasma_plaque_v3 is the recommended alternative for Gel Doc wells and toxoplasma_plaque_v4 stays selectable.
Scored 2026-10-17 on the 153 test fields of the curated 2026-10-15 dataset (spacr.plaque.segment_plaque_image, defaults): object F1 0.816 (95% bootstrap CI 0.778 to 0.848), v4 0.725, v3 0.741, stock cpsam_v2 0.227. On the 81 test fields neither v3 nor v4 saw: v5 0.803, v4 0.661, v3 0.684. AJI is level with v4 (0.737 against 0.744); Gel Doc wells 0.868 against v4 0.901 and v3 0.954.
The final epoch 100 ships by the maintainer’s rule: an earlier checkpoint replaces it only if better on every metric in every domain. Epoch 40 led on validation object F1 (0.766 against 0.750) but trailed on boundary F1 (0.700 against 0.711), AJI (0.661 against 0.667) and Dice (0.857 against 0.869); on test it scored F1 0.823.
Training ran 100 epochs; indices 91 to 99 were resumed from epoch 90 after a power cut, with the same schedule and seeds (optimizer moments restarted at LR 2e-8).
Published as einarolafsson/toxoplasma-plaque-segmentation-cpsam-r7, as cpsam_plaque_r7.
SHA-256 9243917beefbc3d1c8d997d56c1980b811ffef9ee075dac60b099aec715fa8a1.
Toxoplasma Well Detector v3 (YOLO11 Gel Doc, default)¶
Architecture. YOLO11n (fine-tuned from detector v3)
Trained on. 452 reviewed training images; 124 validation images; physical plate and figure groups; 150 epochs; YOLO11n v3 initialization
Measured. Stock was not evaluated in this run; see the named incumbent comparison on the model card.
Metric |
This model |
Stock |
Difference |
|---|---|---|---|
mAP50 |
0.9940 |
not measured |
not measured |
mAP50-95 |
0.8842 |
not measured |
not measured |
Precision |
0.9698 |
not measured |
not measured |
Recall |
0.9979 |
not measured |
not measured |
Evaluation set: detector_v3_validation_figure_clean; version 20261007T205603; 73 scored fields and 191 annotated objects.
The default well detector since 2026-10-12, by the maintainer’s choice. On the 73 clean incumbent-validation images its mAP50-95 is lower than the incumbent’s; on the 3 Gel Doc validation plates it is higher.
Gel Doc plates selected best.pt, so their scores are validation rather than independent test.
Distinct from the YOLO26 checkpoint also historically called detector v4.
Locates wells, not plaques; one class plaque_well (0).
Published as einarolafsson/toxoplasma-plaque-well-detector-yolo11-v4-geldoc, as yolo_welldetect_v4_geldoc.pt.
SHA-256 046861021594256957292cfe9488a84e758077a12dac31fe66b8b4083f8dfa18.