Coral Cover: Chagos #2 2015

Shaded value = in the CV mean (shade deepens with cover) Grey italic = shown for comparison but left out of the CV mean: the gold standard and manual CoralNet annotations (human reference values, not models), Coralscapes (not trusted), and any value that reuses the reference's own data n = frames behind the value — = not run or not applicable (see each region's notes) Site names and method names link out (new tab)

Coral Cover: Chagos #2 2015

49 sites · 1,176 photos

Gold standard = LOF diver transects (no photos). 480 photos manually scored (via CoralNet).

Site	Date	Claude	Gold standard	Manually annotated in CoralNet	CoralNet	DINOv3	Coralscapes v1	Coralscapes v2	CoralSCOP	Mean ± SD of CV	Range
BIPB60	2015-04-14	29.5n=38	42.2	38.8n=38	—	33.4n=38	55.5n=38	57.2n=38	41.9n=38	34.9 ± 6.3	29.5–41.9
BIPB61	2015-04-14	17.0n=14	20.0	17.9n=14	—	22.1n=14	41.0n=14	43.0n=14	32.5n=14	23.9 ± 7.9	17.0–32.5
BIPB62	2015-04-14	33.6n=19	59.6	48.0n=19	—	35.3n=19	57.6n=18	57.5n=18	31.6n=19	33.5 ± 1.9	31.6–35.3
BIPB63	2015-04-15	35.0n=29	56.0	46.9n=29	—	33.3n=29	66.2n=29	66.7n=29	38.4n=29	35.6 ± 2.6	33.3–38.4
BIPB64	2015-04-15	22.8n=21	31.2	22.7n=21	—	26.2n=21	34.3n=21	41.4n=21	33.3n=21	27.4 ± 5.4	22.8–33.3
BIPB65	2015-04-15	30.3n=19	41.4	32.9n=19	—	32.3n=19	52.3n=19	54.9n=19	27.5n=19	30.0 ± 2.4	27.5–32.3
BIPB66	2015-04-16	2.3n=6	7.0	12.5n=6	—	26.1n=6	41.9n=6	40.9n=6	28.8n=6	19.1 ± 14.6	2.3–28.8
BIPB67	2015-04-16	26.1n=41	41.9	36.4n=41	—	33.8n=41	49.3n=41	52.2n=41	29.5n=41	29.8 ± 3.9	26.1–33.8
BIPB68	2015-04-16	36.8n=43	55.0	44.9n=43	—	38.5n=43	71.9n=43	73.8n=43	39.2n=43	38.2 ± 1.2	36.8–39.2
BIPB69	2015-04-16–05-02	39.1n=116	64.7	54.2n=110	56.1n=6	47.4n=116	74.4n=116	75.4n=116	42.6n=116	46.3 ± 7.4	39.1–56.1
BIPB70	2015-04-17–05-01	34.9n=55	60.3	45.4n=28	45.1n=27	40.2n=55	56.6n=55	59.5n=55	35.7n=55	39.0 ± 4.7	34.9–45.1
BIPB72	2015-04-18–05-01	44.3n=75	72.4	50.5n=41	56.1n=34	49.2n=75	64.4n=75	69.0n=75	39.9n=75	47.4 ± 7.0	39.9–56.1
BIPB73	2015-04-18	37.6n=45	50.1	—	45.2n=45	45.3n=45	48.0n=45	50.6n=45	46.9n=45	43.8 ± 4.2	37.6–46.9
BIPB74	2015-04-18–05-01	51.1n=21	62.4	39.3n=8	56.0n=13	48.9n=21	64.5n=21	65.9n=21	47.1n=21	50.8 ± 3.9	47.1–56.0
BIPB75	2015-04-19	32.4n=9	43.9	—	39.2n=9	42.2n=9	51.0n=9	56.5n=9	43.7n=9	39.4 ± 5.0	32.4–43.7
BIPB76	2015-04-19	20.9n=20	35.9	—	29.9n=20	31.5n=20	34.7n=20	39.8n=20	29.0n=20	27.8 ± 4.7	20.9–31.5
BINI77	2015-04-20	25.7n=11	43.0	—	29.8n=11	26.1n=11	38.2n=11	41.1n=11	33.5n=11	28.8 ± 3.6	25.7–33.5
BINI78	2015-04-20	24.8n=16	19.1	—	24.4n=16	29.0n=16	26.9n=16	28.6n=16	28.4n=16	26.6 ± 2.4	24.4–29.0
BINI79	2015-04-20	17.9n=15	31.5	—	19.4n=15	21.6n=15	27.2n=15	25.0n=15	21.2n=15	20.0 ± 1.7	17.9–21.6
BINI80	2015-04-21	29.4n=24	42.8	37.6n=10	26.8n=14	36.9n=24	42.4n=24	45.9n=24	34.9n=24	32.0 ± 4.7	26.8–36.9
BIVB81	2015-04-21	56.7n=5	51.7	—	37.8n=5	44.7n=5	60.9n=5	63.4n=5	49.2n=5	47.1 ± 7.9	37.8–56.7
BIBL82	2015-04-22	28.3n=22	37.9	17.5n=4	27.6n=18	26.1n=22	42.0n=22	50.2n=22	35.8n=22	29.4 ± 4.3	26.1–35.8
BIBL83	2015-04-22	23.8n=15	33.1	20.8n=9	25.6n=6	24.7n=15	42.2n=15	45.2n=15	40.7n=15	28.7 ± 8.0	23.8–40.7
BISA84	2015-04-22	17.5n=7	48.8	25.8n=2	16.7n=5	28.6n=7	35.3n=7	34.3n=7	40.1n=7	25.7 ± 11.0	16.7–40.1
BIBL86	2015-04-23	34.4n=27	47.2	—	31.9n=27	36.0n=27	50.5n=27	49.4n=27	25.1n=27	31.9 ± 4.8	25.1–36.0
BIBL87	2015-04-23	30.3n=35	40.6	—	27.9n=35	34.0n=35	41.5n=35	48.0n=35	29.5n=35	30.4 ± 2.6	27.9–34.0
BISA88	2015-04-23	42.7n=13	56.7	—	45.8n=13	54.1n=13	55.3n=13	59.9n=13	43.4n=13	46.5 ± 5.2	42.7–54.1
BISA91	2015-04-24	40.3n=13	52.4	—	46.7n=13	55.6n=13	58.0n=13	64.8n=13	44.0n=13	46.6 ± 6.5	40.3–55.6
BISP89	2015-04-24	19.9n=13	33.2	—	17.2n=13	21.5n=13	28.6n=13	29.9n=13	27.6n=13	21.6 ± 4.4	17.2–27.6
BISP90	2015-04-24	24.7n=8	39.8	—	27.6n=8	30.8n=8	41.6n=8	43.2n=8	37.9n=8	30.2 ± 5.7	24.7–37.9
BISA92	2015-04-25	18.8n=39	24.5	—	29.4n=39	25.6n=39	33.1n=39	34.3n=39	29.0n=39	25.7 ± 4.9	18.8–29.4
BISA93	2015-04-25	33.0n=27	39.1	—	38.7n=27	26.4n=28	58.3n=26	61.0n=27	26.4n=28	31.1 ± 5.9	26.4–38.7
BISA94	2015-04-25	23.1n=2	40.3	—	47.4n=2	56.7n=2	43.5n=2	63.2n=2	21.4n=2	37.1 ± 17.6	21.4–56.7
BISA97	2015-04-26	51.1n=12	54.4	—	51.1n=12	49.7n=12	67.8n=12	69.3n=12	29.4n=12	45.3 ± 10.6	29.4–51.1
BISP95	2015-04-26	17.8n=26	56.9	—	21.9n=26	20.4n=26	28.0n=26	27.6n=26	23.1n=25	20.8 ± 2.3	17.8–23.1
BISP96	2015-04-26	51.5n=18	86.0	—	55.6n=18	49.3n=18	69.0n=18	72.7n=18	48.2n=18	51.1 ± 3.3	48.2–55.6
BIVB99	2015-04-27	35.4n=11	55.6	—	20.8n=11	24.5n=11	64.2n=11	62.9n=11	18.5n=11	24.8 ± 7.5	18.5–35.4
BISA102	2015-04-28	21.5n=27	30.9	—	31.1n=27	30.1n=27	33.2n=27	34.7n=27	34.4n=27	29.3 ± 5.5	21.5–34.4
BISP100	2015-04-28	31.9n=7	53.9	—	21.7n=7	34.8n=7	42.2n=7	43.2n=7	27.0n=7	28.9 ± 5.8	21.7–34.8
BISP101	2015-04-28	23.0n=13	50.0	—	39.1n=13	32.1n=13	42.2n=13	43.5n=13	32.6n=13	31.7 ± 6.6	23.0–39.1
BISA103	2015-04-29	30.1n=19	32.5	—	28.4n=19	37.2n=19	40.5n=19	44.5n=19	36.1n=19	33.0 ± 4.3	28.4–37.2
BISA104	2015-04-29	35.7n=16	36.2	—	44.2n=16	45.6n=16	46.0n=16	50.6n=16	41.0n=16	41.6 ± 4.4	35.7–45.6
BISA105	2015-04-29	27.8n=19	50.3	—	23.8n=19	36.0n=19	41.8n=19	47.0n=19	23.4n=19	27.8 ± 5.8	23.4–36.0
BIPB110	2015-04-30	29.5n=45	39.3	—	36.6n=45	36.6n=45	44.3n=43	49.6n=43	36.7n=44	34.9 ± 3.6	29.5–36.7
BISA107	2015-04-30	18.5n=12	24.6	—	18.0n=12	20.6n=12	32.1n=12	41.2n=12	21.7n=12	19.7 ± 1.8	18.0–21.7
BIVB108	2015-04-30	11.7n=11	45.7	40.0n=1	25.1n=10	23.9n=11	23.2n=10	18.1n=11	28.2n=11	22.2 ± 7.2	11.7–28.2
BIVB109	2015-04-30	35.0n=33	34.7	36.7n=1	35.0n=32	39.0n=33	67.3n=33	66.1n=33	24.9n=33	33.5 ± 6.0	24.9–39.0
BIPB112	2015-05-02	28.7n=14	45.0	43.6n=14	—	44.3n=14	50.6n=14	55.0n=14	37.7n=14	36.9 ± 7.8	28.7–44.3
BIPB114	2015-05-02–03	20.0n=29	39.6	29.0n=22	34.4n=7	31.7n=29	37.0n=29	38.5n=29	35.2n=29	30.3 ± 7.0	20.0–35.2
All sites (unweighted mean)		29.7	44.1	—	—	35.1	47.3	50.1	33.8	32.8 ± 2.8	29.7–35.1

Notes for Chagos #2 2015 (12)

[a] Value read from cv_method_tables.json (injector output, generated 2026-09-28 22:33); the scope text there gives denominator and frame set.
[b] Gold = LOF in-water diver transect (Mayfield & Dempsey 2026, Oceans 7:1, suppl. S1): a different instrument, site-level only, never within 200 m of the frames.
[c] Anderson's own CoralNet points on the ground-truth frames at this site (30 − WATER − FISH, Unk kept).
[d] CoralNet robot labels (the cover set), same denominator.
[e] DINOv3 point predictions at the same CoralNet points (CRD AI/chagos_cv_2026-09/chagos_dinov3_point_predictions.csv, 2,500 frames).
[f] Coralscapes (v1 EPFL-ECEO/coralscapes-vit-b-dpt; v2 coralscapesv2-dinov3-vitb-lora-dpt) is not trusted (Anderson): shown, never averaged. Area fraction, not a point count.
[g] GENUINELY NOT RUN: only the EDITED website photos were staged for CoralSCOP (Chagos_coralscop_20260920, 2,498 ok); no unedited Chagos CoralSCOP run exists (wide search 2026-09-29).
[h] CoralSCOP = per-frame sum of colony-mask area fractions (zero-filled; overlapping masks double-count), averaged over frames with an ok row. Area index, not a point count: denominator differs from any point-count reference.
[i] Claude arm chagos_claude_arm COMPLETE (2,500/2,500 frames; 2,495 assigned to these 104 site rows): blind calls at 30 CoralNet points on the EDITED website photos, HC+BL / (30 − WATER − FISH), Unk kept.
Mean/Range: Mean ± SD and Range are over the CV cells that enter the mean (CoralNet where independent, DINOv3, CoralSCOP unedited/edited, Claude). Gold standard, frames manually annotated in CoralNet, Coralscapes and same-measurement cells are shown but never averaged (*).
Methods: Claude (Anthropic); DINOv3 (Meta; paper; used here as the frozen backbone of the in-house classifier); Coralscapes v1 and v2 (EPFL-ECEO; dataset paper, code); CoralSCOP (Zheng et al., CVPR 2024); CoralNet.
Sites: names drop the region prefix (hover for the full name) and link to the site's CRD page, or to the regional summary page where the site has none. Date = survey date from the map pins (first–last visit); sites are listed earliest first, undated sites last.

All regions

This trip: what has run, photo basis and photo counts

Trip	Dives	Photos	Reference / transects	Manual points	CoralNet	Claude	DINOv3	Coralscapes v2	Coralscapes v1	CoralSCOP	Table
Chagos #2 2015	57	1,373	LOF GRE in-water transects (suppl. S1)	600 frames / 18,000 pts both trips (480 frames at #2 sites)	1,900 frames robot both trips (695 at #2 sites)	2,500/2,500 EDITED both trips (1,175 at #2 sites) - complete	2,500 EDITED both trips	2,500 EDITED both trips	2,595 EDITED both trips	2,498 ok EDITED both trips (unedited never run)	included