Dom/anonymisation
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

feat: Benchmark de performance baseline - 2.62s/doc moyen, 92% dans objectif

Browse Source
This commit is contained in:
Domi31tls
2026-03-02 10:42:15 +01:00
parent f61e767ee6
commit 30a6ebcc19
108 changed files with 33195 additions and 6 deletions
Download Patch File Download Diff File Expand all files Collapse all files

26
tests/ground_truth/benchmarks/baseline_benchmark.csv Normal file
View File

@@ -0,0 +1,26 @@
pdf,time_s,pii_count
001_simple_unknown_BACTERIO_23018396.pdf,0.38307929039001465,43
002_simple_unknown_bacterio_476_23159413.pdf,0.7698535919189453,47
003_simple_compte_rendu_CRO_23155084.pdf,0.41591382026672363,25
004_simple_anapath_anapath_53_23224186.redacted_raster.pdf,0.3458268642425537,0
005_simple_compte_rendu_CRH_23155836.pdf,0.8738148212432861,140
008_simple_trackare_trackare-14004105-23202435_14004105_23202435.pdf,0.4308145046234131,93
009_simple_compte_rendu_CRO_23051225.pdf,0.47577404975891113,36
010_simple_anapath_ANAPATH_23217289.pdf,0.39705705642700195,54
011_moyen_compte_rendu_CRH_23080179.pdf,1.0042967796325684,46
012_moyen_compte_rendu_CRH_692_23200418.pdf,0.8403730392456055,103
013_moyen_compte_rendu_363_23085243_CRO.pdf,0.94016432762146,160
014_moyen_compte_rendu_CRO_23167029.redacted_raster.pdf,0.4384956359863281,0
015_moyen_unknown_CONSULTATION_ANESTHESISTE_23139653.pdf,0.9846677780151367,25
016_moyen_compte_rendu_CRH_23149905.pdf,1.4508278369903564,242
017_moyen_compte_rendu_CRO_23222062.redacted_raster.pdf,0.4350569248199463,0
018_moyen_compte_rendu_CRH_23042753.pdf,1.9062294960021973,233
019_moyen_compte_rendu_CRO_332_23049003.pdf,1.020752191543579,161
020_moyen_compte_rendu_CRO_23084754.redacted_raster.pdf,0.4804375171661377,0
021_moyen_compte_rendu_CRO_23201117.redacted_raster.pdf,0.31412649154663086,0
022_moyen_compte_rendu_cro2_516_23187028.pdf,0.37198877334594727,29
023_complexe_compte_rendu_CRH_23102610.pdf,4.054161310195923,617
024_complexe_trackare_trackare-17001141-23066188_17001141_23066188.pdf,8.550535917282104,804
025_complexe_trackare_trackare-02016820-23095226_02016820_23095226.pdf,17.83988666534424,1622
026_complexe_trackare_trackare-15000536-23074384_15000536_23074384.pdf,12.040966749191284,1056
027_complexe_trackare_trackare-10027557-23183041_10027557_23183041.pdf,8.782238721847534,859
1 pdf time_s pii_count
2 001_simple_unknown_BACTERIO_23018396.pdf 0.38307929039001465 43
3 002_simple_unknown_bacterio_476_23159413.pdf 0.7698535919189453 47
4 003_simple_compte_rendu_CRO_23155084.pdf 0.41591382026672363 25
5 004_simple_anapath_anapath_53_23224186.redacted_raster.pdf 0.3458268642425537 0
6 005_simple_compte_rendu_CRH_23155836.pdf 0.8738148212432861 140
7 008_simple_trackare_trackare-14004105-23202435_14004105_23202435.pdf 0.4308145046234131 93
8 009_simple_compte_rendu_CRO_23051225.pdf 0.47577404975891113 36
9 010_simple_anapath_ANAPATH_23217289.pdf 0.39705705642700195 54
10 011_moyen_compte_rendu_CRH_23080179.pdf 1.0042967796325684 46
11 012_moyen_compte_rendu_CRH_692_23200418.pdf 0.8403730392456055 103
12 013_moyen_compte_rendu_363_23085243_CRO.pdf 0.94016432762146 160
13 014_moyen_compte_rendu_CRO_23167029.redacted_raster.pdf 0.4384956359863281 0
14 015_moyen_unknown_CONSULTATION_ANESTHESISTE_23139653.pdf 0.9846677780151367 25
15 016_moyen_compte_rendu_CRH_23149905.pdf 1.4508278369903564 242
16 017_moyen_compte_rendu_CRO_23222062.redacted_raster.pdf 0.4350569248199463 0
17 018_moyen_compte_rendu_CRH_23042753.pdf 1.9062294960021973 233
18 019_moyen_compte_rendu_CRO_332_23049003.pdf 1.020752191543579 161
19 020_moyen_compte_rendu_CRO_23084754.redacted_raster.pdf 0.4804375171661377 0
20 021_moyen_compte_rendu_CRO_23201117.redacted_raster.pdf 0.31412649154663086 0
21 022_moyen_compte_rendu_cro2_516_23187028.pdf 0.37198877334594727 29
22 023_complexe_compte_rendu_CRH_23102610.pdf 4.054161310195923 617
23 024_complexe_trackare_trackare-17001141-23066188_17001141_23066188.pdf 8.550535917282104 804
24 025_complexe_trackare_trackare-02016820-23095226_02016820_23095226.pdf 17.83988666534424 1622
25 026_complexe_trackare_trackare-15000536-23074384_15000536_23074384.pdf 12.040966749191284 1056
26 027_complexe_trackare_trackare-10027557-23183041_10027557_23183041.pdf 8.782238721847534 859

151
tests/ground_truth/benchmarks/baseline_benchmark.json Normal file
View File

@@ -0,0 +1,151 @@
{
"date": "2026-03-02T10:41:42.160164",
"batch_date": "2026-03-02T10:29:28.280047",
"configuration": {
"use_ner": true,
"use_vlm": false
},
"statistics": {
"total_documents": 25,
"total_time_s": 65.54734015464783,
"avg_time_s": 2.621893606185913,
"median_time_s": 0.8403730392456055,
"min_time_s": 0.31412649154663086,
"max_time_s": 17.83988666534424,
"stdev_time_s": 4.432960605030657,
"total_pii": 6395,
"avg_pii": 255.8,
"median_pii": 54,
"min_pii": 0,
"max_pii": 1622,
"docs_per_second": 0.3814037295947744,
"pii_per_second": 97.5630740303433
},
"documents": [
{
"pdf": "001_simple_unknown_BACTERIO_23018396.pdf",
"time_s": 0.38307929039001465,
"pii_count": 43
},
{
"pdf": "002_simple_unknown_bacterio_476_23159413.pdf",
"time_s": 0.7698535919189453,
"pii_count": 47
},
{
"pdf": "003_simple_compte_rendu_CRO_23155084.pdf",
"time_s": 0.41591382026672363,
"pii_count": 25
},
{
"pdf": "004_simple_anapath_anapath_53_23224186.redacted_raster.pdf",
"time_s": 0.3458268642425537,
"pii_count": 0
},
{
"pdf": "005_simple_compte_rendu_CRH_23155836.pdf",
"time_s": 0.8738148212432861,
"pii_count": 140
},
{
"pdf": "008_simple_trackare_trackare-14004105-23202435_14004105_23202435.pdf",
"time_s": 0.4308145046234131,
"pii_count": 93
},
{
"pdf": "009_simple_compte_rendu_CRO_23051225.pdf",
"time_s": 0.47577404975891113,
"pii_count": 36
},
{
"pdf": "010_simple_anapath_ANAPATH_23217289.pdf",
"time_s": 0.39705705642700195,
"pii_count": 54
},
{
"pdf": "011_moyen_compte_rendu_CRH_23080179.pdf",
"time_s": 1.0042967796325684,
"pii_count": 46
},
{
"pdf": "012_moyen_compte_rendu_CRH_692_23200418.pdf",
"time_s": 0.8403730392456055,
"pii_count": 103
},
{
"pdf": "013_moyen_compte_rendu_363_23085243_CRO.pdf",
"time_s": 0.94016432762146,
"pii_count": 160
},
{
"pdf": "014_moyen_compte_rendu_CRO_23167029.redacted_raster.pdf",
"time_s": 0.4384956359863281,
"pii_count": 0
},
{
"pdf": "015_moyen_unknown_CONSULTATION_ANESTHESISTE_23139653.pdf",
"time_s": 0.9846677780151367,
"pii_count": 25
},
{
"pdf": "016_moyen_compte_rendu_CRH_23149905.pdf",
"time_s": 1.4508278369903564,
"pii_count": 242
},
{
"pdf": "017_moyen_compte_rendu_CRO_23222062.redacted_raster.pdf",
"time_s": 0.4350569248199463,
"pii_count": 0
},
{
"pdf": "018_moyen_compte_rendu_CRH_23042753.pdf",
"time_s": 1.9062294960021973,
"pii_count": 233
},
{
"pdf": "019_moyen_compte_rendu_CRO_332_23049003.pdf",
"time_s": 1.020752191543579,
"pii_count": 161
},
{
"pdf": "020_moyen_compte_rendu_CRO_23084754.redacted_raster.pdf",
"time_s": 0.4804375171661377,
"pii_count": 0
},
{
"pdf": "021_moyen_compte_rendu_CRO_23201117.redacted_raster.pdf",
"time_s": 0.31412649154663086,
"pii_count": 0
},
{
"pdf": "022_moyen_compte_rendu_cro2_516_23187028.pdf",
"time_s": 0.37198877334594727,
"pii_count": 29
},
{
"pdf": "023_complexe_compte_rendu_CRH_23102610.pdf",
"time_s": 4.054161310195923,
"pii_count": 617
},
{
"pdf": "024_complexe_trackare_trackare-17001141-23066188_17001141_23066188.pdf",
"time_s": 8.550535917282104,
"pii_count": 804
},
{
"pdf": "025_complexe_trackare_trackare-02016820-23095226_02016820_23095226.pdf",
"time_s": 17.83988666534424,
"pii_count": 1622
},
{
"pdf": "026_complexe_trackare_trackare-15000536-23074384_15000536_23074384.pdf",
"time_s": 12.040966749191284,
"pii_count": 1056
},
{
"pdf": "027_complexe_trackare_trackare-10027557-23183041_10027557_23183041.pdf",
"time_s": 8.782238721847534,
"pii_count": 859
}
]
}