- Rating
- 3(+1)
- Confidence
- 3
- Informativeness
- 2
- Presentation
- 2
- Soundness
- 2
Read review
Paper summary
This paper proposes GPS, a hard-positive learning framework that addresses the insufficient modeling of identity-preserving cross-view variations in aerial-ground person re-identification. GPS first employs GCPS to generate cross-view candidates with varying strengths, then uses HPPS to select the hardest candidate within a prescribed difficulty range while satisfying identity constraints. Finally, DPEF fuses supported evidence from the original and selected candidate features to improve identity representation learning.
Critical strengths
- The paper models aerial-ground viewpoint differences as changes in the spatial distribution, relative contribution, and visibility of local identity evidence, which is intuitive and well aligned with practical AG-ReID scenarios.
- Using real same-identity cross-view pairs as references, the method generates identity-consistent candidates with controllable additional difficulty, making it more targeted than unconstrained data augmentation.
- The appendix provides detailed method parameters, training configurations, and evaluation protocols, offering substantial information for reproducibility.
- The synthesis, selection, and fusion modules are used only during training. Inference still requires a single image encoding followed by cosine-distance retrieval, introducing no additional inference overhead.
Critical weaknesses
- About GCPS
Appendix B.1 explicitly defines the coordinate prior zi, but this prior is essentially a fixed grid function and does not incorporate actual camera angles or spatial coordinates. It implicitly assumes that the person is roughly centered and that the body axis is vertical; tilted poses or detection-box offsets may violate these assumptions.
Merging and vanishing treat the min–max normalized token norm as a local response and use it to compute support. Although intuitively reasonable, this assumption is not directly validated. Moreover, the vanishing risk function tends to select tokens with lower positional support and weaker responses. This creates some tension with its motivation of modeling cases where “distinctive details become invisible” (under this proxy, salient cues with stronger responses are instead more likely to be retained).
- About HPPS
HPPS computes Δk using a global representation h that combines the class token and pooled patch information. Local edits affecting only a small region may be diluted during feature aggregation, and the paper does not verify whether this global difficulty measure accurately reflects the strength of local edits.
The design of HPPS is intuitively reasonable, but the paper provides no statistics or analysis of its actual selection behavior, such as the candidate acceptance rate or the proportion of samples with no accepted candidate. It is therefore difficult to determine whether the module indeed selects candidates that are moderately difficult and identity-consistent.
- About experiments
All comprehensive ablations are conducted on the synthetic CARGO dataset, which is insufficient to establish the independent contribution and stability of each component under real-world capture conditions.
The paper lacks a generic augmentation baseline under matched conditions. For example, if random token perturbations combined with HPPS and DPEF yield comparable improvements, the main gains may come from appropriate hard samples and additional supervision rather than the geometry-guided design.
Questions for authors
-
How robust is the fixed coordinate prior in GCPS to tilted poses and off-center detections? Could the authors provide analyses under these conditions?
-
Is there empirical evidence that the min–max normalized token norm reliably reflects local identity evidence or support? In particular, since the vanishing risk function favors tokens with lower positional support and weaker responses, how does it model the disappearance of distinctive, highly salient identity cues?
-
How well does the global difficulty measure Δk, computed from the class token and pooled patch information, capture changes in matching difficulty caused by edits affecting only a small number of tokens? Could the authors validate its sensitivity to localized changes?
-
Could the authors report statistics on the actual behavior of HPPS, such as the candidate acceptance rate and the proportion of samples with no accepted candidate, as well as the distributions of selected candidate strengths and Δk?
-
Could the authors provide component ablations on at least one dataset with real captured images to verify that the individual contributions and stability observed on synthetic CARGO data generalize to real-world capture conditions?
-
How does GPS compare with generic feature-space augmentation under matched conditions? For example, could the authors evaluate random token perturbation, masking, or mixing using the same number of candidates, HPPS, DPEF, and training objectives, to isolate the contribution of the geometry-guided design?
8 comments
Some low-response patches on the body still contain subtle appearance details. In the figure, some low-response patches appear near the hands, waist, or feet or in the background.
In addition, we compared three response definitions on AG-ReID v2.
- Position prior only. Use positional support alone to guide the edits.
- Quantile-clipped normalization. Clip the tails of the within-image token-norm distribution at quantile bounds, then rescale the clipped norms.
- Min–max token norm (GPS). Apply min–max normalization to the raw norms of the current patch sequence.
The table reports mAP / Rank-1 (%) on the four AG-ReID v2 protocols.
Table L: Comparison of token-response definitions on AG-ReID v2.
| Response definition | A→C | C→A | A→W | W→A |
|---|---|---|---|---|
| Position prior only | 81.72 / 87.69 | 80.68 / 87.08 | 85.55 / 90.76 | 81.54 / 87.61 |
| Quantile-clipped normalization | 82.35 / 88.03 | 81.22 / 87.63 | 86.18 / 91.49 | 82.37 / 88.59 |
| Min–max token norm (GPS) | 84.75 / 89.90 | 83.58 / 89.01 | 88.29 / 93.35 | 85.11 / 90.98 |
The min–max token norm improves both mAP and Rank-1 over both alternatives on all four protocols, with mAP gains of 2.11 to 2.74 percentage points over quantile-clipped normalization. These results support using the within-image min–max norm as the local response for patch editing.
Vanishing conservatively attenuates local details after compression and merging. It selects a small set of regions with weaker token responses to simulate the partial loss of local evidence. Suppressing the most discriminative regions can make a hard positive unreliable as a same-identity sample. We will revise the phrase “distinctive details” in the paper to avoid implying that vanishing suppresses salient cues. We also tried prioritizing high-response tokens for vanishing, which made candidates too difficult and led to the loss of salient identity cues. On all four AG-ReID v2 protocols, prioritizing high-response tokens gives lower mAP and Rank-1 than the original GPS. The table reports mAP / Rank-1 (%).
Table M: Comparison of vanishing high- and low-response tokens on AG-ReID v2.
| Vanishing setting | A→C | C→A | A→W | W→A |
|---|---|---|---|---|
| Prioritize high-response tokens | 81.16 / 87.27 | 79.84 / 86.86 | 84.24 / 90.40 | 81.77 / 88.55 |
| Original GPS (prioritize low-response tokens) | 84.75 / 89.90 | 83.58 / 89.01 | 88.29 / 93.35 | 85.11 / 90.98 |
In addition, with compression and merging fixed, adding vanishing on the real-capture LAGPeR G→AG protocol raises mAP / Rank-1 from 25.91 / 33.55 to 27.85 / 36.18 (A5 below). The ablation gain also supports preferentially weakening a small set of low-response tokens after compression and merging, with positional support guiding the selection.
Q3: (a) HPPS computes Δk using a global representation h that combines the class token and pooled patch information. Local edits affecting only a small region may be diluted during feature aggregation, and the paper does not verify whether this global difficulty measure accurately reflects the strength of local edits.
(b) How well does the global difficulty measure Δk, computed from the class token and pooled patch information, capture changes in matching difficulty caused by edits affecting only a small number of tokens? Could the authors validate its sensitivity to localized changes?
A3: Retrieval uses global descriptors, so local edits aim to make same-identity matching harder at the level of the global representation. Changes in global similarity show whether the edits affect retrieval. We used two CARGO diagnostics to examine edit strength and extent.
First, we compared unedited features with candidates generated at five strengths for the same 960 CARGO sources. We scored the candidates generated by full GPS. The table reports mean in units of .
Table N: Mean added difficulty by synthesis strength on CARGO.
| Generation level ℓ | Mean |
|---|---|
| No edit | ≈0 |
| 0.3500 | 3.096539 |
| 0.5125 | 3.862922 |
| 0.6750 | 4.645124 |
| 0.8375 | 5.443020 |
| 1.0000 | 6.256495 |
The unedited control has a mean of approximately zero. With editing, the mean is positive at all five strengths and increases as the edits get stronger. This positive relationship indicates that stronger patch edits increase global matching difficulty, yielding harder same-identity positives.
Second, we held the CLS token and donor fixed and kept the actual GCPS edit residuals at a specified number of source patch positions. The multiplier controls the magnitude of the retained residuals. The table reports mean in units of .
Table O: Mean added difficulty by edited-token count and residual scale on CARGO.
| Edited tokens out of 128 | α = 0.25 | α = 0.50 | α = 1.00 |
|---|---|---|---|
| 1 | 0.180 | 0.189 | 0.218 |
| 4 | 0.491 | 0.782 | 0.916 |
| 16 | 2.416 | 3.151 | 6.732 |
| 32 | 6.316 | 10.067 | 21.663 |
| 64 | 16.724 | 34.345 | 101.650 |
| 128 | 71.691 | 260.571 | 1105.382 |
Within each column, at fixed , the difficulty increment increases as more tokens are edited. Within each row, at a fixed number of edited tokens, stronger edits also increase the difficulty increment. The global difficulty measure thus captures the extent and strength of local edits.
Q4: (a) The design of HPPS is intuitively reasonable, but the paper provides no statistics or analysis of its actual selection behavior, such as the candidate acceptance rate or the proportion of samples with no accepted candidate. It is therefore difficult to determine whether the module indeed selects candidates that are moderately difficult and identity-consistent.
(b) Could the authors report statistics on the actual behavior of HPPS, such as the candidate acceptance rate and the proportion of samples with no accepted candidate, as well as the distributions of selected candidate strengths and Δk?
A4: We analyzed HPPS selection behavior on CARGO. In the training logs, every batch contains 64 original benchmark images, and 100% of sources have an observed same-identity cross-view reference. During the active synthesis stage, 75% of eligible sources enter GCPS. GCPS generates five candidates for each selected source, and HPPS accepts at most one. The recorded mean source-level acceptance is 79.88%, corresponding to estimated coverage of 75% × 79.88% = 59.91%, or about 38 of 64 original inputs receiving an accepted synthetic-feature update. These updates add cross-view evidence beyond the original observations used for training.
In the candidate-level analysis, approximately 25.5501% of candidates passed all gates, 77.3456% of participating sources had at least one accepted candidate, and 22.6544% fell back to their original features. Among the selected candidates, Δₖ had a mean of 0.0050357, a median of 0.0048971, and a 5th–95th percentile interval of [0.0040995, 0.0064198] within the default [0.004, 0.007] band.
We also tallied the shares of accepted synthetic samples at each nominal strength, with 953 recorded selections in total.
Table P: Distribution of HPPS selections by nominal synthesis strength on CARGO.
| Nominal strength ℓ | Selections | Share of selections |
|---|---|---|
| 0.3500 | 1 | 0.1049% |
| 0.5125 | 31 | 3.2529% |
| 0.6750 | 181 | 18.9927% |
| 0.8375 | 378 | 39.6642% |
| 1.0000 | 362 | 37.9853% |
The two highest nominal strengths account for 77.65% of the selections. This pattern is consistent with the Q3 results, where the two lowest strengths produce mean Δₖ values below the 0.004 lower bound, and with HPPS choosing the hardest candidate that satisfies the identity and realized difficulty checks.
Q5: (a) All comprehensive ablations are conducted on the synthetic CARGO dataset, which is insufficient to establish the independent contribution and stability of each component under real-world capture conditions.
(b) Could the authors provide component ablations on at least one dataset with real captured images to verify that the individual contributions and stability observed on synthetic CARGO data generalize to real-world capture conditions?
A5: We ran the operator, module, and training objective ablations on all four datasets while preparing the paper. Because of space constraints, Table 2 in the main text presents only the CARGO ablations. We will add the complete ablation results for all four datasets to the appendix. The tables below report all LAGPeR and AG-ReID v1 protocols and two AG-ReID v2 protocols. The controls follow the settings in Table 2 and Appendix C.3. Entries are mAP / Rank-1 (%).
LAGPeR
Table Q: Ablations on LAGPeR.
| Ablation setting | A→G | G→A | G→AG |
|---|---|---|---|
| Operators | |||
| No C, M, or V | 29.86 / 42.09 | 33.06 / 34.37 | 21.45 / 27.91 |
| C | 32.61 / 44.85 | 35.90 / 37.98 | 24.56 / 31.94 |
| M | 32.37 / 45.04 | 35.50 / 37.46 | 23.87 / 31.55 |
| V | 31.44 / 43.63 | 34.69 / 36.57 | 23.41 / 30.60 |
| C + M | 33.94 / 46.85 | 37.04 / 39.30 | 25.91 / 33.55 |
| C + V | 34.14 / 46.88 | 37.39 / 39.72 | 26.20 / 33.75 |
| M + V | 33.60 / 46.29 | 37.06 / 39.30 | 25.51 / 33.22 |
| C + M + V | 35.67 / 48.56 | 39.02 / 41.63 | 27.85 / 36.18 |
| Modules | |||
| Clean control | 29.20 / 40.77 | 32.39 / 34.24 | 21.13 / 27.58 |
| GCPS | 34.53 / 46.29 | 37.71 / 40.15 | 26.35 / 34.41 |
| GCPS + HPPS | 35.37 / 48.16 | 38.71 / 41.33 | 27.52 / 35.72 |
| GCPS + DPEF | 35.17 / 48.13 | 38.35 / 41.00 | 27.39 / 35.65 |
| GCPS + HPPS + DPEF | 35.67 / 48.56 | 39.02 / 41.63 | 27.85 / 36.18 |
| Objectives | |||
| No GVPS or OSIC | 34.17 / 46.95 | 37.41 / 39.82 | 26.22 / 33.85 |
| GVPS | 34.75 / 47.14 | 38.18 / 40.64 | 27.02 / 35.26 |
| OSIC | 34.97 / 47.70 | 38.33 / 40.68 | 27.24 / 35.33 |
| GVPS + OSIC | 35.67 / 48.56 | 39.02 / 41.63 | 27.85 / 36.18 |
We sincerely thank you for your thorough review and suggestions.
A1:
(a)
First, person ReID typically follows pedestrian detection. The detector provides a bounding box, and the ReID model receives a crop around the person. This makes the person's location and body layout relatively consistent in the model input. We also observed this pattern across several real-capture benchmarks. We manually reviewed a stratified random sample of 360 training crops. A crop was compatible if the principal body axis was within 30° of vertical and the horizontal and vertical offsets of the center of the body's extent from the crop center were at most 20% of the crop width and height, respectively. Most training crops were compatible with the prior, with particularly high compatibility rates on real-capture benchmarks.
Table K: Manual audit of crop layouts.
Second, explicit viewpoint projection and scale modeling require richer per-observation information, such as camera intrinsics, viewing angle, and camera pose or height. GPS is designed to use the person crops and aerial–ground view labels already provided by standard AG-ReID benchmarks without requiring additional annotations. In future work, we will further explore using actual camera geometry for finer geometric modeling.
Finally, the fixed prior provides a shared coordinate reference and serves only as soft supervision. The locations of actual edits depend on the input image's patch response strengths and vary across samples. The crop center serves as a positional reference, and patch features can match across grid locations. This lets cues from tilted or off-center pedestrians participate in matching where they appear. For example, if a shifted detection box places a shoulder cue to one side of the crop, its token response features can still establish a correspondence with a donor patch.
(b)
We applied random rotation, horizontal translation, and vertical translation separately, then retrained and evaluated GPS, ViSA, and GSAlign under matched conditions. Relative to the original results, GPS scores dropped by about 1–2 percentage points, compared with about 3–5 points for ViSA and GSAlign. GPS still achieved the highest mAP and Rank-1 on both AG-ReID v2 protocols under each matched perturbation. The smaller drops indicate greater robustness to tilted poses and off-center crops. We believe the geometric prior and GCPS's use of input token responses help the method adapt to these changes.
Table K1: Rotation and translation robustness on AG-ReID v2 (mAP / Rank-1, %).
A2: To illustrate how local responses inform support, we visualized patch-token responses. Please refer to Figure R1 in §4.9 of the latest version of the paper. The top row shows the normalized token-response heatmaps, their overlays, and the 26 highest-response tokens (Top26). High responses are concentrated mainly on the pedestrian's body.