QAConv-QA: Quality-Aware Query-Adaptive Conv

• The paper introduces QAConv-QA, a pixel-level matching mechanism that adaptively generates convolution filters and enforces bidirectional constraints to robustly address clothes-changing re-identification.
• It integrates dual-branch feature extraction with multi-modal attention fusion to combine RGB and parsing cues, producing discriminative and clothing-invariant representations.
• Experiments on PRCC and LTCC benchmarks show significant performance gains, confirming the effectiveness of spatial quality weighting and bidirectional matching in noisy environments.

Quality-Aware Query-Adaptive Convolution (QAConv-QA) is a pixel-level matching framework developed to address the problem of clothes-changing person re-identification (CC-ReID), where traditional appearance cues become unreliable due to significant intra-identity clothing variations. QAConv-QA extends earlier query-adaptive convolution methods by incorporating spatially-varying quality weighting and enforcing bidirectional matching constraints, thereby enhancing robustness to clothing variation and background noise. The module is integrated into the broader QA-ReID architecture, which employs dual-branch feature extraction and multi-modal attention fusion to jointly leverage global appearance and clothing-invariant structural cues (Wang et al., 27 Jan 2026).

1. Integration within Dual-Branch Feature Extraction

QAConv-QA operates on fused feature maps within QA-ReID. Each image $I \in \mathbb{R}^{3 \times H' \times W'}$ is passed through two parallel ResNet-50 branches truncated at stage-3: an RGB branch producing $$F_{\text{rgb}} \in \mathbb{R}^{C \times H \times W}$$, and a parsing branch generating a binary body mask $M_{\text{body}} \in \{0,1\}^{H'\times W'}$ and extracting $$F_{\text{par}} \in \mathbb{R}^{C \times H \times W}$$. A multi-modal attention module computes channel- and spatial-wise attention weights $\omega \in [0,1]^{C \times H \times W}$ across the concatenated features $[F_{\text{rgb}}; F_{\text{par}}]$. The fused features are obtained as:

• $$F_{\text{mix}} = \omega \odot F_{\text{rgb}} + (1-\omega) \odot F_{\text{par}}$$
• $$F = F_{\text{rgb}} + F_{\text{par}} + F_{\text{mix}}$$
• $F_{\text{fuse}} =$ 1×1 convolution($F$), resulting in $$F_{\text{rgb}} \in \mathbb{R}^{C \times H \times W}$$0.

QAConv-QA is applied to pairs of these fused features from probe and gallery images to compute a fine-grained similarity score.

2. Query-Adaptive Convolution Filter Generation

QAConv-QA dynamically generates a local convolution kernel at each spatial position of the probe's fused feature map. Specifically, for the probe $$F_{\text{rgb}} \in \mathbb{R}^{C \times H \times W}$$1 of dimension $$F_{\text{rgb}} \in \mathbb{R}^{C \times H \times W}$$2, a learnable 1×1 convolution $$F_{\text{rgb}} \in \mathbb{R}^{C \times H \times W}$$3 is applied to yield $$F_{\text{rgb}} \in \mathbb{R}^{C \times H \times W}$$4. The slice at each $$F_{\text{rgb}} \in \mathbb{R}^{C \times H \times W}$$5 is reshaped into a kernel $$F_{\text{rgb}} \in \mathbb{R}^{C \times H \times W}$$6, which is convolved over the gallery feature $$F_{\text{rgb}} \in \mathbb{R}^{C \times H \times W}$$7 at the corresponding position. Stacking these responses for all $$F_{\text{rgb}} \in \mathbb{R}^{C \times H \times W}$$8 results in a raw pixel-to-pixel similarity tensor.

This procedure allows for localized, dynamic, query-conditioned matching, in contrast to global or fixed-filter approaches.

3. Pixel-Level Importance Weighting

Not all spatial locations in the feature map contribute equally to identity discrimination, especially under clothing changes and background clutter. QAConv-QA computes a spatial quality (or importance) map $$F_{\text{rgb}} \in \mathbb{R}^{C \times H \times W}$$9, derived from the downsampled body mask:

• The original mask $M_{\text{body}} \in \{0,1\}^{H'\times W'}$0 is pooled over $M_{\text{body}} \in \{0,1\}^{H'\times W'}$1 patches:

$M_{\text{body}} \in \{0,1\}^{H'\times W'}$2

• A spatial softmax normalizes these coarse mask sums:

$M_{\text{body}} \in \{0,1\}^{H'\times W'}$3

During matching, the score for a pixel pair $M_{\text{body}} \in \{0,1\}^{H'\times W'}$4 is weighted as follows:

$M_{\text{body}} \in \{0,1\}^{H'\times W'}$5

where $M_{\text{body}} \in \{0,1\}^{H'\times W'}$6 denotes cosine similarity.

This weighting suppresses contributions from background and unreliable regions, emphasizing body structure over appearance.

4. Bidirectional Consistency Constraints

Single-directional similarity matching can yield high scores for spurious correspondences. QAConv-QA introduces bidirectional normalization to enforce mutual matching. For each pair $M_{\text{body}} \in \{0,1\}^{H'\times W'}$7:

• Forward-normalized score:

$M_{\text{body}} \in \{0,1\}^{H'\times W'}$8

• Reverse-normalized and combined:

$M_{\text{body}} \in \{0,1\}^{H'\times W'}$9

Only pairs that are the mutually most likely correspondences yield large contributions. A bidirectional global max-pooling (Bi-GMP) operation further aggregates the similarity map to a scalar, followed by batch normalization, an MLP, and sigmoid to produce a final match probability $$F_{\text{par}} \in \mathbb{R}^{C \times H \times W}$$0.

This mechanism explicitly penalizes one-way mismatches, increasing reliability under appearance changes.

5. Matching Workflow

A probe-gallery matching instance proceeds as follows:

1. Extract $$F_{\text{par}} \in \mathbb{R}^{C \times H \times W}$$1, $$F_{\text{par}} \in \mathbb{R}^{C \times H \times W}$$2 for both probe and gallery via respective branches.
2. Fuse with attention $$F_{\text{par}} \in \mathbb{R}^{C \times H \times W}$$3 to generate $$F_{\text{par}} \in \mathbb{R}^{C \times H \times W}$$4, $$F_{\text{par}} \in \mathbb{R}^{C \times H \times W}$$5.
3. Generate $$F_{\text{par}} \in \mathbb{R}^{C \times H \times W}$$6 query-adaptive filters $$F_{\text{par}} \in \mathbb{R}^{C \times H \times W}$$7 from $$F_{\text{par}} \in \mathbb{R}^{C \times H \times W}$$8.
4. Convolve each $$F_{\text{par}} \in \mathbb{R}^{C \times H \times W}$$9 over $\omega \in [0,1]^{C \times H \times W}$0 to obtain raw responses.
5. Compute pixel-level quality maps $\omega \in [0,1]^{C \times H \times W}$1, $\omega \in [0,1]^{C \times H \times W}$2 and weight the responses.
6. Apply bidirectional normalization as above.
7. Aggregate with Bi-GMP, apply batch normalization, MLP, and sigmoid to yield the scalar similarity $\omega \in [0,1]^{C \times H \times W}$3.

6. Training Objectives and Loss Formulation

The QA-ReID framework containing QAConv-QA is trained end-to-end using three loss components:

• Cross-entropy identity losses on global pooled features of each branch:

$\omega \in [0,1]^{C \times H \times W}$4

• Triplet losses on both branches:

$\omega \in [0,1]^{C \times H \times W}$5

• Pairwise matching loss for QAConv-QA:

$\omega \in [0,1]^{C \times H \times W}$6

where $\omega \in [0,1]^{C \times H \times W}$7 if images $\omega \in [0,1]^{C \times H \times W}$8 and $\omega \in [0,1]^{C \times H \times W}$9 share the same ID.

The overall loss is:

$[F_{\text{rgb}}; F_{\text{par}}]$0

This joint objective encourages discriminative, identity-preserving representation in both color/texture and structure, and robust pixel-level matching under clothing changes.

7. Performance Analysis and Impact

Empirical results substantiate the effectiveness of QAConv-QA’s mechanisms in the CC-ReID setting. Experiments on PRCC and LTCC benchmarks show:

• Baseline fusion with plain QAConv (no pixel weighting, no bidirectional constraints) attains 61.4% Top-1 on PRCC and 41.1% on LTCC.
• Adding pixel weighting alone increases performance to 63.0% / 41.8%.
• Incorporating bidirectional matching singly yields 63.7% / 42.4% (value for LTCC inferred by juxtaposition, this suggests bidirectional constraints alone moderately help).
• The full QAConv-QA with both enhancements achieves 64.1% Top-1 on PRCC and 42.9% on LTCC, representing a $[F_{\text{rgb}}; F_{\text{par}}]$1 absolute gain on PRCC over QAConv and $[F_{\text{rgb}}; F_{\text{par}}]$2 over the strongest existing competitor.

The explicit use of spatial quality weighting and reciprocal normalization distinguishes QAConv-QA from conventional global-matchers and the original QAConv, leading to increased robustness against large clothing changes, background interference, and unreliable regions (Wang et al., 27 Jan 2026).