Authors: Bessie Dominguez-Dager, Miguel Cazorla, Vincent Lepetit
Organizations: Institute for Computing Research, University of Alicante, Alicante, Spain · LIGM, École Nationale des Ponts et Chaussées, IP Paris, Univ Gustave Eiffel, CNRS, France
Cloth-changing person re-identification (CC-ReID) is still very challenging, and no method clearly dominates: methods that lead on one benchmark collapse on the others, because they rest on different identity cues. We show that a general-purpose vision-language model (VLM), used zero-shot as a judge for reranking a short candidate list, beats the expert retrieval models it reranks by a wide and consistent margin. On the four scenarios of CHIRLA, a long-term CC-ReID benchmark that no publicly released Re-ID model is trained on, reranking the candidates of two frozen experts raises rank-1 on every scenario, by 12.9 points on average over the best expert and 9.4 over the best naive ensemble of the two, with no component fine-tuned on the target data. Merging both experts' shortlists matches or beats reranking the stronger expert's shortlist alone, at lower cost. Introducing a VLM does, however, come at a cost: almost all of the pipeline's FLOPs are spent on it. We therefore borrow a criterion from ensembling and invoke the VLM judge only where the two experts disagree at rank 1. This threshold-free gate removes 40% of the calls, and a comparable share of total compute, with no significant drop in rank-1, trading 1.8 to 3.8 points of mAP. Judging remains the dominant cost, but on a benchmark where dedicated experts stall, we find a frozen general-purpose model to be the strongest judge of identity. The code is available at https://github.com/bdager/JudgeReID.
Figures & tables
Figure 1: State-of-the-art CC-ReID experts make different mistakes : We consider here DIFFER and SapiensID on CHIRLA [ 22 ] , a challenging multi-camera scenario. Left: two queries on which the experts disagree, with their top-3 candidates; green marks correct identities, red incorrect ones. Each model succeeds where the other fails. Middle: an oracle selecting the better model per query gains 13.5 rank-1 points over the best expert (dashed). Right: queries split by top-1 agreement and by which model is correct. Agreement is a reliable correctness signal, and in the disagreement column only 13% of queries have no correct candidate in either top-6 list. In this paper, we introduce and evaluate different approaches to rerank the union of both shortlists with a VLM judge, on every query or, to save compute, only where the two disagree.
Gated
Correct
Oracle (or.)
Pool recall
Scenario
(% all)
(% agree)
R-1 m⋆→ or.
(% gated)
LT
74.6
64.2
35.5→50.1
83.9
MC-LT
65.2
70.9
43.4→61.8
94.0
MC
54.9
85.5
61.3→74.7
90.2
Reapp.
47.0
79.1
65.5→75.5
95.2
Avg.
60.4
74.9
51.4→65.5
90.8
Table 1: Gate analysis. Gated : the experts’ R-1 identities differ, so the judge is invoked. Correct : agreeing queries whose shared R-1 is right. Oracle : the better expert m⋆ against a per-query oracle. Pool recall : gated queries whose GT identity is in the merged P=16 pool; the judge’s R-1 ceiling.
Method
LT
MC-LT
MC
Reapp.
Avg.
Δ R-1 [95% CI]
R-1
mAP
R-1
mAP
R-1
mAP
R-1
mAP
R-1
mAP
DIFFER [ 21 ]
30.89
31.20
42.56
48.51
52.67
53.79
54.15
55.45
45.07
47.24
−6.4[−12.6,+0.2]
SapiensID [ 12 ]
35.50
35.01
43.44
46.63
61.28
57.94
65.50
61.22
51.43
50.20
ref.
Voting (Borda) [ 66 ]
38.63
36.17
45.81
50.27
60.33
59.27
62.84
61.00
51.90
51.68
+0.5[−3.2,+4.3]
RRF [ 28 ]
39.60
37.41
49.49
52.14
61.06
60.47
63.20
62.31
53.34
53.08
+1.9[−1.8,+5.8]
Score Fusion [ 27 ]
40.95
38.31
48.60
51.78
64.47
61.75
65.93
63.90
54.99
53.94
+3.6[+0.3,+7.0]
Table 2: Main results on CHIRLA . Our pipeline against the two experts and standard ensembling. Per column, the best and second-best results, in bold and underlined , respectively. Δ is the Avg. R-1 difference relative to SapiensID (ref.) with its 95% confidence interval (CI) from a paired bootstrap over identities; ∗ marks p<0.05 under an exact sign-flip test ( Section A.5 ).
Figure 2: Accuracy against compute as P grows. R-1 is averaged over the four scenarios and compute is the total FLOPs to run all four benchmarks once. All three pipelines sit above the SapiensID embedding-only baseline (dashed). The gated pipeline traces the same accuracy range at a fraction of the cost, since it calls the judge only where the experts disagree.
Table 5
Scenario
(DIFFER, SapiensID)
(SapiensID, DIFFER)
Δ
Agreement % (chance)
R-1
mAP
R-1
mAP
R-1
mAP
@1
@ kH
@ P
LT
44.97
39.59
45.94
39.88
+0.98
+0.29
69.6 (31.2)
42.9 (4.9)
31.2 (2.3)
MC-LT
56.35
54.43
58.50
55.11
+2.15
+0.68
69.2 (33.6)
49.7 (6.1)
39.7 (3.6)
MC
73.25
64.87
73.21
64.93
−0.04
+0.06
79.4 (33.2)
51.5 (5.6)
39.2 (3.0)
Reapp.
73.72
65.28
73.43
65.01
−0.30
−0.27
82.1 (34.2)
55.4 (5.9)
41.1 (2.9)
Table 5: Sensitivity to ρ at P=6 , with two orderings of C(q) . Δ is the second order minus the first. Agreement is the fraction of gated queries on which the two orders return the same identity sequence, at rank 1, over the retained head ( kH=3 ), and over the whole pool, whereas R-1 and mAP are over the full query set. In parentheses, chance agreement under random permutations of the same pool.
Method
LT
MC-LT
MC
Reapp.
Format (%)
R-1 ↑
mAP ↑
R-1 ↑
mAP ↑
R-1 ↑
mAP ↑
R-1 ↑
mAP ↑
Corr. ↓
Unrec. ↓
Qwen3.6-35B-A3B
43.96
38.70
55.58
53.52
70.92
63.80
72.13
64.33
6.79
1.76
Qwen3.6-27B
44.97
39.59
56.35
54.43
73.25
64.87
73.72
65.28
0.23
0.12
Table 6: Results for two VLMs at P=6 , gated pipeline. Format reports how often the response was not valid, with Corr. meaning repaired by the correction prompt, and Unrec. still invalid after the maximum of attempts, computed over the four Re-ID scenarios.
Figure 3: Qualitative results on two gated queries. Query in gray , green marks a correct identity and red an incorrect one. For each query the upper row is the merged pool from DIFFER and SapiensID in the order shown to the judge, and the lower row is the order the judge returns. Above, the judge lifts four correct candidates to the head. Below, it demotes the three correct candidates.
Appendix figures & tables7 assets
Supplementary material from the paper’s appendix.
Appendix
Scenario
Subsets
Query
Gallery
Cams.
All
Closed
IDs
Imgs.
IDs
LT
6
4903
3692
10
368
13
6
MC-LT
9
2207
1134
11
252
19
7
MC
9
4454
4311
19
305
19
7
Reapp.
9
1251
1251
21
258
21
4
Appendix
Table 7: CHIRLA statistics. Subsets counts the train_ k /test_ k pairs the protocol evaluates, excluding subset 0 . Sizes are summed over the pairs; identity and camera counts are unions over them. Closed counts the queries whose identity is not a distractor and appears in the corresponding gallery, and the gap to all is what the closed-set protocol discards.
Method
LT
MC-LT
MC
Reapp.
Avg.
PFLOPs
R-1
mAP
R-1
mAP
R-1
mAP
R-1
mAP
R-1
mAP
DIFFER [ 21 ]
30.89
31.20
42.56
48.51
52.67
53.79
54.15
55.45
45.07
47.24
1.9
SapiensID [ 12 ]
35.50
35.01
43.44
46.63
61.28
57.94
65.50
61.22
51.43
50.20
0.6
Voting (Borda) [ 66 ]
38.63
36.17
45.81
50.27
60.33
59.27
62.84
61.00
51.90
51.68
2.4
RRF [ 28 ]
39.60
37.41
49.49
52.14
61.06
60.47
63.20
62.31
53.34
53.08
2.4
Score Fusion [ 27 ]
40.95
38.31
48.60
51.78
64.47
61.75
65.93
63.90
54.99
53.94
2.4
Appendix
Table 8: CHIRLA Results over P . VLM: Qwen3.6-27B . In bold best result per column and methods grouped. PFLOPs is the total forward compute to run all four benchmarks once, summed over scenarios: backbone passes over the closed-set queries and gallery, plus 2NactD for the D tokens the VLM processes ( Nact=27 B active parameters).
Comparison
R-1
mAP
Δ [95% CI]
p
k
Δ [95% CI]
p
k
Against SapiensID
Union + VLM (gated) P=16 vs. SapiensID
+12.9[+9.3,+16.8]
< 0.001
18
+9.5[+7.1,+11.7]
< 0.001
20
Union + VLM (gated) P=6 vs. SapiensID
+10.6[+7.0,+14.4]
< 0.001
19
+5.8[+3.4,+8.1]
< 0.001
20
Score Fusion vs. SapiensID
+3.6[+0.3,+7.0]
0.084
19
+3.7[+1.5,+5.9]
0.012
21
RRF vs. SapiensID
+1.9[−1.8,+5.8]
0.426
19
+2.9[+0.3,+5.3]
0.078
21
Appendix
Table 9: Paired uncertainty for every comparison in Section 4 . Δ is the first method minus the second, in points; for Binding and Removal rows it is the change from adding or removing the named prompt component. Values are on the Avg. column, except the prompt ablations, which were run on Reapp. only. Intervals are 95% from the identity-level paired bootstrap. p is the exact sign-flip test over identities and k the number of identities whose share of Δ is non-zero, so the smallest attainable p is 2/2k . p<0.05 in bold, uncorrected for multiple comparisons.
Scenario
P=6
P=10
P=16
P=20
P=30
P=40
SapiensID + VLM
LT
0.2 / 0.1
0.2 / 0.1
5.3 / 1.5
8.9 / 1.6
20.9 / 1.7
32.0 / 2.6
MC-LT
0.2 / 0.2
0.1 / 0.0
3.4 / 0.1
4.3 / 0.0
7.1 / 0.1
9.6 / 0.4
MC
0.1 / 0.1
0.1 / 0.1
4.6 / 0.0
7.2 / 0.0
7.5 / 0.4
22.7 / 1.4
Reapp.
0.1 / 0.4
0.4 / 0.0
3.7 / 0.0
6.8 / 0.0
16.2 / 0.3
18.9 / 0.5
Union + VLM (gated)
Appendix
Table 10: Output-contract failures as a function of P . Each cell is corrected / unrecovered , as a percentage of the queries sent to the judge. Judge: Qwen3.6-27B . See Section A.6 .
Figure 4: A query and its P=6 pool. The images are sent as P+1 separate entries after the instructions, in the order ρ . Each crop carries its role in a white banner burned into the pixels, Query or the candidate number, which is the Label intervention of Table 4 . Without it the role survives only in the surrounding text, and the judge has to infer which crop is which from position alone.
Removed component
R-1
Δ
mAP
Δ
None, full prompt
73.72
65.28
Attribute list, system message
72.47
−1.25
64.46
−0.82
Attribute list, user message
72.74
−0.98
64.57
−0.71
Clothing prohibition and escape hatch
73.12
−0.60
64.82
−0.46
Escape hatch only
72.05
−1.68
64.44
−0.84
Step-by-step directive
73.56
−0.16
64.90
−0.38
Appendix
Table 11: Semantic removals from the full prompt, CHIRLA (Reappearance), P=6 . Each row removes one component from the full prompt, the last row of Table 4 , which is repeated here for reference. Δ is measured against it. No removal is significant under the sign-flip test ( Table 9 ).
Tail rule
LT
MC-LT
MC
Reapp.
Fixed tail: DIFFER
38.32
54.23
63.09
63.61
Fixed tail: SapiensID
38.51
52.95
63.05
64.39
Majority tail
credit DIFFER
39.49
54.64
64.77
64.99
credit SapiensID
39.59
54.35
64.69
65.27
credit own-rank (tie DIFFER)
39.52
54.62
64.81
65.11
Appendix
Table 12: Tail rule ablation. mAP (%) at P=6 with head size kH=k=3 , per-subset averages, SapiensID as the agreement source. Judge: Qwen3.6-27B . See Section A.8 for the rules.