Files
aimsight-labeling/auto/analyses/FLORIDA_2026-10-11T0945_2026-10-22T0315/analysis.md
T
maximeandClaude Opus 5.5 ddb17c2189 auto: labels requalified under the 5-noise rules (only what stands out from the noise): 32 analyses refined, the 6 paused periods verified, the long labels requalified with the fixes the period analysts found; 1153 -> 961 labels
- 32 analyses: each label kept, rewritten or removed by its analyst (section Requalification of the report; mm magnitudes where the targets have them)
- SNCF01 x3, WILDSTRUBEL 2026-09-01 -> 2026-10-06 x3: verified from their existing labels under the new rules
- long labels (no analysis): auto/analyses/long-labels/<STATION>/; NORDEND-169 and -195 run to 07/11, so NORDEND-283 and -284 (their continuation) are removed
- reading-rules.md, stations.md: the citations of removed or rewritten labels point to their successor, or to [expert]

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-11-17 23:58:00 +01:00

100 lines
12 KiB
Markdown
Raw Blame History

This file contains ambiguous Unicode characters
This file contains Unicode characters that might be confused with other characters. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.
# Labels: FLORIDA, 2026-10-11T09:45Z → 2026-10-22T03:15Z (UTC)
Verification of the 23 existing labels: 23 kept, 0 corrected, 0 removed, 5 added. The work was cut short by the usage limit: the checks listed under "Not checked" were not done.
Requalification of 2026-11-17 under the new rules (only what stands out from the noise of the corrected curves): of the 28 labels, 23 kept unchanged, 0 rewritten, 5 removed (FLORIDA-30, 33, 36, 38, 43). Details in section 2 and in "Requalification".
Weather: there is no `weather/` folder for FLORIDA, so no weather is available.
## Verdict per old id
- FLORIDA-23 kept: r1_r is rejected at 10:00 and there is no pose at that instant (inventory).
- FLORIDA-24, 25, 26, 28, 31, 33, 36, 38, 44 kept. In the hourly pose (metrics.py), pan, tilt and scale follow the housing temperature. They start and come back within one or two measurements of the bounds (pan near -0.033 at night and -0.05 to -0.056 at the peak; scale from 1.0005 to 1.0015). (Requalification: 33, 36, 38 removed, see section 2.)
- FLORIDA-41 kept, with a doubt. In the hourly pose, the tilt starts falling and the scale rising from 08:00, before the gap at 09:00 (tilt 0.0098 at 07:00, 0.0057 at 08:00, 0.0026 at 10:00). The cycle may therefore start around 07:45 rather than 10:45. This was not checked at the native step.
- FLORIDA-27, 29, 32, 34, 35, 40, 42 kept: their bounds match the gaps listed by the inventory exactly. The claims about other stations were not re-checked.
- FLORIDA-30 kept (not re-checked at the native step). (Requalification: removed, see section 2.)
- FLORIDA-37 and FLORIDA-39 kept: they cover all the rejections of p1_r, r1_r, Sellette 4, Sellette 5 and Pont Ernen on the 18th and 19th, and the missing poses.
- FLORIDA-43 kept: Sellette 5 is rejected at 05:30 and 05:45 and is back at 06:00. (Requalification: removed, see section 2.)
- FLORIDA-45 kept: the last rejections are r1_r at 18:15 and Pont Ernen at 18:00. From 18:30 the pose is back, at the cold level. After that, Sellette 4's corr_x stays a little lower (about 0.075 against 0.11 the night before), which is one or two noise widths: not labeled.
- Spanning FLORIDA-1 and FLORIDA-14: consistent with the rejections of Sellette 6 and Massif 2 in the range (inventory).
## 1. My labels (added)
### 2026-10-12T00:30Z → 01:00Z, 01:45Z → 03:00Z, 03:45Z → 04:30Z, 06:15Z → 06:45Z, 22:15Z → 22:45Z
- **Phenomenon of the catalogue**: closest is "a target latches onto another spot", here applied to a reference; no exact match.
- **Precedents**: JUNGFRAU-7 and FLORIDA-51 (a reference poorly seen at night distorts the pose: Visibility). Not followed, because here p1_r's error_value does not rise.
- **Family**: Obstruction, because the cause is a local point beside one spot, not the whole scene. The nature of the point is not established: medium confidence.
- **Measured**: the jumps are at 00:45, 02:00, 02:45, 04:00 and 06:30 on the 12th, then 22:30. Sellette 4 corr_x rises by about 0.15 to 0.2 px (about 1.1 to 1.7 mm), which is 5 to 10 noise widths. In the rigid fit on the targets (`overview.py --rigid` from 00:30), p1_r moves by about -0.1 to -0.19 px in dx while the targets stay still. p1_r's error_value stays flat (0.38 to 0.46). Requalification: all five above 5 noises (FLORIDA-74 +10.3, 75 +6.4, 76 +7.0, 77 +6.3 in y, 78 +7.6, Sellette 4, local rolling level).
- **Images**: the fixed-frame thumbnails at 01:00, 02:00, 03:00 and 04:00 show a small bright point at the upper left of p1_r at 01:00 and 03:00 (no jump), absent at 02:00 and 04:00 (jump). Copied to `p1_r_point.png`.
- **Confidence**: medium.
## 2. What I left without a label
- Isolated deviations of 3 to 4 noise widths (14/10 21:00, 21/10 21:15–22:45, and a few others): too small.
- The night of 21–22: the pose sits at a colder level that matches the temperature (tilt +0.02 and scale 1.0002 at 16–17 °C, the same relation as on the other nights). The corrected curves are flat apart from the small offset of Sellette 4 noted above.
### Removed by the requalification of 2026-11-17
Noise = median of the differences between two consecutive measurements over the whole range (Sellette 4: 0.0198 px in x, 0.0162 px in y; Sellette 5: 0.0212 px in x, 0.0184 px in y). Massif 2 and Sellette 6 are left aside in the judgement of a thermal or movement label: their deviations belong to their month-long Target issue labels (FLORIDA-14, FLORIDA-1), whose kept measurements jump and change level.
- **FLORIDA-30**, 2026-10-15T04:00Z → 09:00Z, Displacement: inside the label, the largest deviation from the level of the 3 h before is Sellette 5 corr_x -4.9 noises (at 08:45, already inside the thermal cycle FLORIDA-31), Sellette 4 corr_x -3.2. Before the thermal start (04:00–07:45), the level of Sellette 5 corr_x moves by only -0.9 noise from the night before. The return it describes, read night against night (14th 00–05 against 17th 00–05), is 4.3 noises for Sellette 5 corr_x (about 1.2 mm) and 3.4 for Sellette 4 corr_x: slow, spread over three days and across the outage of the 15–16th, under the threshold. Removed.
- **FLORIDA-33**, 2026-10-16T10:45Z → 21:00Z, Thermal displacement: against the night level (17th 00–05; the night of the 16th is in the outage), at most Sellette 5 corr_y +4.5 noises, Sellette 5 corr_x -3.6, Sellette 4 +3.2; the label itself said "the corrected curves show almost nothing". Compensated thermal movement of the camera: removed.
- **FLORIDA-36**, 2026-10-17T09:45Z → 23:00Z, Thermal displacement: against the night level (17th 00–05), at most Sellette 5 corr_y +4.0 noises, Sellette 4 corr_x +2.6, Sellette 5 corr_x -2.4. Massif 2 corr_y reaches +11.9 noises at 21:15, but in the evening and as jumps of its kept measurements (its Target issue FLORIDA-14), not a daily cycle. Compensated thermal: removed.
- **FLORIDA-38**, 2026-10-18T08:45Z → 19:00Z, Thermal displacement: against the night level (18th 00–05), the curves of Sellette 4 and Sellette 5 stay within ±5 noises all day except two isolated single measurements: Sellette 4 corr_x +8.7 noises at 10:15 and Sellette 4 corr_y -5.2 at 10:30. The third largest deviation of Sellette 4 corr_x is -2.8: no cycle in the corrected curves; the label itself said "show almost nothing". Compensated thermal: removed. The one-measurement spikes of 10:00–10:45 (rolling level, noise of the range: none for Sellette 5 in x; in the local noise of a 14 h window, Sellette 4 corr_x +15.7 at 10:15, corr_y -14.6 at 10:30, Sellette 5 corr_y -7.2 at 10:00 and +5.9 at 10:45) stand out but are not the thermal cycle; under the instruction not to look for new events, they are left unlabeled here and flagged for the next analysis of this range.
- **FLORIDA-43**, 2026-10-20T05:15Z → 06:00Z, Obstruction: missing points only. Sellette 5 is rejected at 05:30 and 05:45 (its curve is missing); Sellette 4 moves at most 1.3 noises, Sellette 6 and the references do not move (no range in the rolling run). Removed.
### Kept, borderline
- FLORIDA-24 (thermal, 11th): Sellette 5 corr_x -5.5 noises at 11:00 against the night of the 11th; Sellette 4 corr_y at -3 to -6 noises all afternoon against the night of the 12th (3-point running median -6.1 at 14:30). Just above: kept.
- FLORIDA-44 (thermal, 20th): Sellette 5 corr_x at -3 to -4 noises all day, -5.7 at 11:45 and -6 at 15:15; Sellette 4 corr_y +5.2 at 13:15. Just above: kept.
- FLORIDA-41 (thermal, 19th): Sellette 5 corr_x -5 to -7 noises at 13:00–13:30 and -9 at 21:15 (against the night of the 20th; the night of the 19th is under FLORIDA-39). Kept.
- FLORIDA-23: Sellette 5 corr_x +5.3 noises (rolling level), +7.9 against the level of the range, at 10:00. Kept.
- FLORIDA-29 (No data) holds one delivered measurement (02:15) between its two empty instants; the station delivers nothing at the two others: kept as it is (bounds not moved).
## Requalification
Commands, from the repository root, outputs in `work/req/`:
- `python3 tools/scripts/ranges.py data/FLORIDA/ --columns corr_x,corr_y --k 5 --start 2026-10-11T09:45Z --end 2026-10-22T03:15Z` → `ranges_k5.txt` (level of the range).
- `python3 tools/scripts/ranges.py data/FLORIDA/ --columns corr_x,corr_y --k 5 --rolling 4 --sensors "Sellette 4,Sellette 5,Massif 2,Sellette 6" --start 2026-10-11T09:45Z --end 2026-10-22T03:15Z` → `ranges_k5_roll4.txt` (local level, noise of the range); the same on 11/10 03:00–16:00 and 17/10 22:00–18/10 12:00 (local noise).
- `python3 <folder>/scripts/requalify.py <folder> 3` → `dev_level3h.txt`: each label, largest deviation from the level of the 3 h before, px and mm.
- `python3 <folder>/scripts/night.py <folder>` → `thermal_night.txt`: thermal labels against the night level 00–05 of the day.
- `python3 <folder>/scripts/smooth.py <start> <end> <night start> <night end>`: measurement by measurement, in noises, for 24, 33, 36, 38, 41, 44.
- No data: every label checked to contain no delivered instant for any sensor (FLORIDA-29: one, 02:15, between its two gaps).
| id | family | largest deviation (Sellette 4 / 5, own noises) | verdict |
|---|---|---|---|
| 23 | Visibility | S5 x +5.3 (rolling), +7.9 (range level) | kept |
| 24 | Thermal | S5 x -5.5; S4 y -6.1 (running median) | kept |
| 74 | Obstruction | S4 x +10.3, S5 x +6.8 | kept |
| 75 | Obstruction | S4 x +6.4 | kept |
| 76 | Obstruction | S4 x +7.0 | kept |
| 77 | Obstruction | S4 y +6.3, S4 x +5.4 | kept |
| 25 | Thermal | S5 x -11.2 (3rd largest -5.6) | kept |
| 78 | Obstruction | S4 x +7.6, S5 x +5.0 | kept |
| 26 | Thermal | S5 x -7.1 (3rd -6.1) | kept |
| 27, 29, 32, 34, 35, 40, 42 | No data | no measurement from any sensor | kept |
| 28 | Thermal | S5 x -6.6 (3rd -5.8) | kept |
| 30 | Displacement | S5 x -4.9; night to night 4.3 | removed |
| 31 | Thermal | S5 x -6.5 (3rd -6.3), S4 x -5.7 | kept |
| 33 | Thermal | S5 y +4.5 | removed |
| 36 | Thermal | S5 y +4.0 | removed |
| 37 | Visibility | S5 x -5.6 (rolling); S4 y -9.1 (local noise) | kept |
| 38 | Thermal | no cycle; single spikes S4 x +8.7, S4 y -5.2 | removed |
| 39 | Visibility | S5 x +12.7, S4 y +12.3 | kept |
| 41 | Thermal | S5 x -8.7, -7 at 13:15 | kept |
| 43 | Obstruction | missing points only; S4 1.3 | removed |
| 44 | Thermal | S5 x -5.7, S4 y +5.2 | kept |
| 45 | Visibility | S5 x +18.2, S4 x +15.7 | kept |
Kept 23, rewritten 0, removed 5.
## 3. The documentation
- The `--rigid` mode of `overview.py` was decisive: it showed that the reference moves, not the targets.
- Outside the range: FLORIDA-46 says "the camera does not move", but at 03:30 on the 22nd the pose jumps and Sellette 5 jumps for one measurement too.
- Requalification: section 1 of `labeling-rules.md` was clear. Not said: which night is "the night level" when the night before is in an outage or under another label (FLORIDA-33, 41); I took the next calm night. And whether a thermal label whose only crossings are isolated single measurements (FLORIDA-38) holds: I judged it does not, since the cycle is absent from the corrected curves.
- The removal of FLORIDA-30 leaves the shift "taken on the 9th" (labeled before this range) without its return label; that label is outside this range and was not re-read.
## Not checked (cut short)
- FLORIDA-41 start at the native step.
- The other stations for the No data labels.
## Difficulties
- The usage limit cut the verification short.
## Sources read
instructions/analyst.md, tools/GUIDE.md, tools/DATA.md, labeling/families.md, labeling-rules.md, reading-rules.md, stations.md, labeling/auto/FLORIDA.jsonl (labels around the range), precedents.py, inventory, overview (with rigid), metrics, days, ranges, scripts/native.py, scripts/uncovered.py, thumbnails of 12/10. Requalification: labeling-rules.md section 1, ranges.py help, data.py (load), work/requalify_input.jsonl, work/spanning.jsonl, work/inventory.txt.