CLIDD: Cross-Layer Independent Deformable Description for Efficient and Discriminative Local Feature Representation

Haodi Yao; Fenghua He; Ning Hao; Yao Su

cs.CVJan 14, 2026

CLIDD: Cross-Layer Independent Deformable Description for Efficient and Discriminative Local Feature Representation

Authors: Haodi Yao, Fenghua He, Ning Hao, Yao Su

Organizations: School of Astronautics, Harbin Institute of Technology, China · State Key Laboratory of General Artificial Intelligence, Beijing Institute for General Artificial Intelligence (BIGAI), China

Abstract

Robust local feature representations are essential for spatial intelligence tasks such as robot navigation and augmented reality. Establishing reliable correspondences requires descriptors that provide both high discriminative power and computational efficiency. To address this, we introduce Cross-Layer Independent Deformable Description (CLIDD), a method that achieves superior discriminability by sampling directly from independent feature hierarchies. This approach utilizes learnable offsets to capture fine-grained structural details across scales, without ever constructing a unified dense feature map, eliminating the memory and computation overhead of building a full-scale representation for a sparse sampling process. The per-keypoint per-layer offsets further break interpolation-induced correlation, preserving descriptor discriminability even at high keypoint densities. To ensure real-time performance, we implement a hardware-aware kernel fusion strategy that maximizes inference throughput. Furthermore, we develop a scalable framework that integrates lightweight architectures with a training protocol leveraging both metric learning and knowledge distillation. This scheme generates a wide spectrum of model variants optimized for diverse deployment constraints. Extensive evaluations demonstrate that our approach achieves superior matching accuracy and exceptional computational efficiency simultaneously. Furthermore, our high-performance configuration outperforms a broad range of current state-of-the-art methods while exceeding 200 FPS on edge devices. These results demonstrate that CLIDD delivers high-precision local feature matching with minimal computational overhead, providing a robust and scalable solution for real-time spatial intelligence tasks.

Figures & tables

Panel of Figure 1 — Figure 1: Structural comparison of local descriptor extraction strategies. (a) Sub-Scale Vanilla aggregates multi-scale features into a single low-resolution dense map for sparse sampling. (b) Full-Scale Vanilla constructs a high-resolution, unified feature field to improve precision at the cost of increased memory overhead. (c) Full-Scale Sparse Deformable Descriptor Head (SDDH)introduces a sparse deformable sampling head on top of the dense, full-scale feature map. (d) Our method employs Cross-Layer Independent Deformable Description (CLIDD) to perform sparse sampling directly from multiple independent feature layers. This strategy generates highly discriminative descriptors while completely bypassing the construction of unified, dense feature maps.

Figure 2: Deformable sampling offsets of our U128 model on three independent feature maps at 1/2-level, 1/8-level, and 1/32-level. All offsets are projected and visualized at the original image scale for comparison.

Figure 3: Precision-efficiency comparison on resource-constrained devices. This plot illustrates the relationship between relative pose estimation accuracy (AUC@10 ∘ ) on the MegaDepth-1500 benchmark and inference speed (FPS) measured on an edge platform. Our CLIDD-based models consistently occupy the upper-right quadrant, outperforming their corresponding state-of-the-art baselines in both matching precision and processing speed across model scales.

Figure 4: Mechanism of Cross-Layer Independent Deformable Description. This framework facilitates sparse feature extraction across multiple hierarchies without relying on a unified dense feature map. The Cross-Layer Predictor utilizes spatially aligned embeddings to determine level-specific sampling offsets for each feature resolution. Guided by these offsets, the Layer-Independent Sampler performs decoupled feature retrieval from each layer, where $M$ denotes the number of deformable sampling points per scale. This mechanism allows the model to capture both fine-grained geometry and semantic context simultaneously. The independently sampled features are then aggregated to produce a discriminative descriptor. While illustrated with three scales, this design is inherently scalable and generalizes to standard deformable sampling in single-layer configurations.

Figure 5: Model Architecture. Our lightweight framework generates multi-scale representations using standard convolutional layers and ResNet blocks. The notation

C_{i}

for

i\in\{1,2,3\}

specifies the output channel counts for each feature block, while

C_{\mathrm{det}}

and

C_{\mathrm{desc}}

denote the dimensions for keypoint detection and final description. Within the description head, the total dimension of concatenated sampled features is defined as

C_{\mathrm{sum}}=C_{1}+C_{2}+C_{3}

. Symbols

\textcircled{\scalebox{0.8}{+}}

and

\textcircled{\scalebox{0.7}{C}}

indicate element-wise addition and concatenation operations, respectively. The term

r_{i}

for

i\in\{2,3\}

determines the number of ResNet blocks within each specific stage. Following extraction, a post-processing module transforms the dense detection heatmap into sparse keypoints. By utilizing our Cross-Layer Independent Deformable Description method, the architecture produces high-quality descriptors without generating dense, high-resolution feature maps, thereby simultaneously enhancing both descriptive precision and computational efficiency.

Model	Network Configuration								Params (MP)
Model	$C_{1}$	$C_{2}$	$C_{3}$	$r_{2}$	$r_{3}$	$C_{\mathrm{det}}$	$M$	$C_{\mathrm{desc}}$	Backbone	Detect	Desc	Total
A48(Atom)	4	4	4	1	1	4	4	48	0.00123	0.00036	0.003	0.004
N64(Nano)	8	8	8	1	1	8	8	64	0.00448	0.00109	0.014	0.019
T64(Tiny)	8	16	24	1	1	8	8	64	0.015	0.00128	0.027	0.043
S64(Small)	8	24	32	1	1	8	16	64	0.026	0.00141	0.072	0.100
M64(Medium)	16	32	48	1	1	8	16	64	0.058	0.00167	0.108	0.168

Table 1: Network configurations. We denote the models as A/N/T/S/M/L/G/E/U to indicate the model size.

Model	$w_{\mathrm{DS}}$	$w_{\mathrm{OP}}$	$w_{\mathrm{US}}$
A48	0.05	1	1
N64	0.1	1	1
T64	0.5	1	1
S64	1	0	1
M64	1	0	1
L64	1	0	1

Table 2: Loss weighting configurations. The table details the coefficients

w_{\mathrm{DS}}

w_{\mathrm{OP}}

, and

w_{\mathrm{US}}

assigned to each model architecture during training.

	Method	Dim	MP	GFLOPs	FPS
Standard	SuperPoint ( DeTone et al., 2018 )	256	1.301	26.11	124.2
	DISK ( Tyszkiewicz et al., 2020 )	128	1.092	98.97	26.9
	ALIKED-T16 ( Zhao et al., 2023 )	64	0.192	1.37	83.3
	ALIKED-N16 ( Zhao et al., 2023 )	128	0.677	4.05	47.7
	ALIKED-N32 ( Zhao et al., 2023 )	128	0.980	4.62	46.1
	AWDesc-T16 ( Wang et al., 2023 )	128	0.172	4.50	180.4

Table 3: Computation Resources Comparison. Our methods achieve superior throughput and resource efficiency for edge applications. The best results are marked as bold .

	Method	MHA
	Method	@1	@3	@5
Compact	ALIKED-T16 ( Zhao et al., 2023 )	50.74	83.89	90.56
	XFeat ( Potje et al., 2024 )	47.04	80.19	89.07
	EdgePoint ( Yao et al., 2024 )	47.96	78.89	87.04
	EdgePoint2-T48 ( Yao et al., 2025 )	50.19	81.48	88.89
	EdgePoint2-M64 ( Yao et al., 2025 )	53.70	82.96	89.63

Table 4: Homography estimation on HPatches. The top three best results in each sector are highlighted in red , green and blue .

	Method	MegaDepth-1500			ScanNet-1500			IMC2022
	Method	AUC@5 ∘	AUC@10 ∘	AUC@20 ∘	AUC@5 ∘	AUC@10 ∘	AUC@20 ∘	Private	Public
Compact	ALIKED-T16 ( Zhao et al., 2023 )	58.46	70.89	79.77	14.10	27.18	40.54	0.542	0.559
	XFeat ( Potje et al., 2024 )	43.61	56.78	67.38	12.14	25.08	39.72	0.489	0.505
	EdgePoint ( Yao et al., 2024 )	42.92	55.72	66.04	11.31	23.46	36.59	0.436	0.460
	EdgePoint2-T48 ( Yao et al., 2025 )	48.71	60.94	70.67	13.72	27.20	42.01	0.550	0.564
	EdgePoint2-M64 ( Yao et al., 2025 )	53.20	65.17	74.05	15.28	29.95	44.89	0.608	0.615

Table 5: Relative pose estimation on MegaDepth-1500, ScanNet-1500 and IMC2022. The top three best results of each sector are marked as red , green and blue .

Figure 6: Qualitative results on MegaDepth-1500. We compare ALIKED-N32, AWDesc-CA and DeDoDe-G against our U128 model.

	Method	Aachen Day-Night v1.1						InLoc
		Day			Night			DUC1			DUC2
		0.25 $m$	0.5 $m$	5 $m$	0.25 $m$	0.5 $m$	5 $m$	0.25 $m$	0.5 $m$	5 $m$	0.25 $m$	0.5 $m$	5 $m$
		2 ∘	5 ∘	10 ∘	2 ∘	5 ∘	10 ∘	2 ∘	5 ∘	10 ∘	2 ∘	5 ∘	10 ∘
Compact	ALIKED-T16 ( Zhao et al., 2023 )	89.1	95.0	98.2	77.5	91.1	99.0	33.8	49.0	59.1	33.6	49.6	57.3
	XFeat ( Potje et al., 2024 )	86.0	93.9	97.6	70.7	85.9	97.9	36.4	54.0	64.1	30.5	44.3	57.3
	EdgePoint ( Yao et al., 2024 )	86.9	93.2	97.6	66.0	85.9	95.3	28.3	43.4	53.0	22.9	37.4	49.6

Table 6: Visual localization on Aachen Day-Night v1.1 and InLoc. The top three best results of each sector are marked as red , green and blue .

Model	Predictor	Sampler	MegaDepth-1500			ScanNet-1500
Model	Predictor	Sampler	AUC@5 ∘	AUC@10 ∘	AUC@20 ∘	AUC@5 ∘	AUC@10 ∘	AUC@20 ∘
A48	Single	Independent	43.99	56.20	65.89	12.41	25.19	38.92
	Cross	Independent	44.98	57.13	66.41	12.62	25.35	39.23
	Cross	Shared	43.20	55.22	64.72	12.15	24.41	37.82
U128	Single	Independent	63.17	75.51	84.14	18.69	35.91	52.74
	Cross	Independent	64.75	77.15	85.77	18.94	36.46	53.66

Table 7: Ablation study of specific structural configurations. These experiments compare single-layer versus cross-layer aggregation for offset prediction and shared versus layer-independent strategies for feature sampling.

Model	Description Architecture	FPS	MegaDepth-1500			ScanNet-1500
Model	Description Architecture	FPS	AUC@5 ∘	AUC@10 ∘	AUC@20 ∘	AUC@5 ∘	AUC@10 ∘	AUC@20 ∘
A48	Sub-Scale Vanilla ( Yao et al., 2025 )	845.3	25.28	36.39	47.46	9.18	18.89	30.77
	Full-Scale Vanilla ( Edstedt et al., 2024a )	420.2	13.47	21.15	30.53	5.12	12.06	21.90
	Full-Scale SDDH ( Zhao et al., 2023 )	162.4	42.56	54.27	63.93	11.42	23.39	36.66
	Ours	881.1	44.98	57.13	66.41	12.62	25.35	39.23
U128	Sub-Scale Vanilla ( Yao et al., 2025 )	170.2	60.82	73.07	81.94	18.28	35.00	51.09

Table 8: Comparison with dense description architectures. The experimental results evaluate our sparse extraction strategy against several dense fusion description method.

ALIKED Model	Description Architecture	MegaDepth-1500			ScanNet-1500
ALIKED Model	Description Architecture	AUC@5 ∘	AUC@10 ∘	AUC@20 ∘	AUC@5 ∘	AUC@10 ∘	AUC@20 ∘
T16	SDDH ( Zhao et al., 2023 )	58.46	70.89	79.77	14.10	27.18	40.54
T16	CLIDD	58.73	71.51	80.48	15.07	29.49	44.17
N16	SDDH ( Zhao et al., 2023 )	57.62	70.00	79.05	13.16	25.90	39.90
N16	CLIDD	59.24	71.46	80.31	15.05	28.89	43.36
N32	SDDH ( Zhao et al., 2023 )	61.10	73.56	82.16	12.72	24.65	37.05

Table 9: Comparison of description architectures on frozen ALIKED models. The ALIKED backbone and detection branch are kept frozen, while the original SDDH is replaced with CLIDD and only the CLIDD parameters are optimized.

Model	$w_{\mathrm{DS}}$	$w_{\mathrm{OP}}$	MegaDepth-1500			ScanNet-1500
Model	$w_{\mathrm{DS}}$	$w_{\mathrm{OP}}$	AUC@5 ∘	AUC@10 ∘	AUC@20 ∘	AUC@5 ∘	AUC@10 ∘	AUC@20 ∘
A48	0	1	43.98	55.93	65.17	11.86	23.91	37.09
	0.05	1	44.98	57.13	66.41	12.62	25.35	39.23
	0.1	1	42.30	54.93	65.25	11.16	23.37	37.56
	1	0	33.87	46.65	57.99	9.86	21.12	34.05
N64	0	1	52.71	64.21	72.80	13.97	27.83	42.01

Table 10: Ablation of descriptor loss weighting. Performance comparison across different weight balances for the DualSoftmax loss and the Orthogonal-Procrustes loss.

Model	Teacher Description	MegaDepth-1500			ScanNet-1500
Model	Teacher Description	AUC@5 ∘	AUC@10 ∘	AUC@20 ∘	AUC@5 ∘	AUC@10 ∘	AUC@20 ∘
A48	ALIKED-N32	44.98	57.13	66.41	12.62	25.35	39.23
A48	Ours-U128	46.43	58.53	67.87	12.56	25.03	38.93
N64	ALIKED-N32	53.91	65.81	75.12	14.88	29.34	44.35
N64	Ours-U128	54.06	65.79	74.55	14.97	29.25	44.04
T64	ALIKED-N32	54.26	66.74	76.16	15.33	31.03	47.37

Table 11: Ablation of teacher model for descriptor distillation. Comparison of descriptor performance with ALIKED-N32 vs. our U128 model as the distillation teacher.

Keypoint Prior		MegaDepth-1500			ScanNet-1500
Detection	Description	AUC@5 ∘	AUC@10 ∘	AUC@20 ∘	AUC@5 ∘	AUC@10 ∘	AUC@20 ∘
ALIKED-N32	ALIKED-N32	64.75	77.15	85.77	18.94	36.46	53.66
ALIKED-N32	SuperPoint	64.85	77.03	85.47	19.00	35.94	53.19
SuperPoint	ALIKED-N32	60.13	72.90	82.15	18.89	36.62	54.31
SuperPoint	SuperPoint	59.57	72.67	82.14	18.68	35.43	52.20

Table 12: Ablation on keypoint position prior. Performance comparison when substituting ALIKED-N32 with SuperPoint for detection supervision and descriptor sampling locations independently.

Figure 7: Throughput comparison across various model scales and keypoint counts. The results visualize the inference efficiency with and without kernel fusion. Both the number of sampling points and the FPS results are presented on a

\log_{2}

scale.

Method	Configuration		MegaDepth-1500			ScanNet-1500			IMC2022
Method	Max $N$	NMS	AUC@5 ∘	AUC@10 ∘	AUC@20 ∘	AUC@5 ∘	AUC@10 ∘	AUC@20 ∘	Private	Public
A48	1024	$\checkmark$	39.08	51.52	61.76	12.39	24.47	37.90	0.402	0.417
	4096	$\checkmark$	44.98	57.13	66.41	12.62	25.35	39.23	0.473	0.492
	4096	$\times$	48.84	61.06	70.32	14.45	28.03	42.31	0.493	0.511
	10000	$\times$	51.96	63.65	72.26	15.37	29.86	44.31	0.530	0.548
N64	1024	$\checkmark$	48.26	60.69	70.38	14.50	28.25	42.51	0.515	0.520

Table 13: Evaluation of descriptor discriminability under varying spatial densities. Comparison of matching performance across different keypoint limits and NMS configurations for multiple model scales.

Figure 8: Performance under varying keypoint density on relative pose estimation. Results compare U128 against ALIKED-N32, DeDoDe-B, DeDoDe-G, and RDD.

Figure 9: Spatial derivative visualization of dense descriptor maps for ALIKED, DeDoDe-G, RDD, and four CLIDD model variants. The descriptor map spatial derivative reveals local correlation patterns. RDD exhibits blurred textures with grid-like artifacts from interpolation. CLIDD models produce sharp and uniform differential textures across all model scales.

Figure 10: Dense keypoint matching results with 10,000 keypoints and without NMS. We compare our lightweight models A48 and N64 against our largest U128.

Panel of Figure 12 — Figure 12: Analysis of learned sampling offsets under scale, rotation, homography, and perspective transformations. For the scale transformation, sampling offsets are computed from the original image and its 150% enlarged version, with all points visualized at a common scale to enable direct comparison. All sampling points are displayed at the original image resolution.

Local Feature Matching

Local feature matching finds corresponding points or regions across images to support geometric estimation, localization, reconstruction, and recognition.

See all 36 papers

Oct 15, 2026

CLAD: Keypoint Matching Using Global Line Correspondences

Traditional keypoint descriptors, such as SIFT (Scale-Invariant Feature Transform) and ORB (Oriented FAST and Rotated BRIEF), are computed from local image patches. They may be unreliable in scenes with repetitive patterns or weak textures, where patches at different locations can look similar. We propose the Contextual Line-Assisted Descriptor (CLAD), a keypoint matching method that addresses this problem by using global geometric context from line correspondences. CLAD represents each keypoint with a vector whose elements indicate its proximity to matched lines. It can therefore use the keypoint's spatial relationships to the lines in the scene. We use HOME (Hough-space One-dimensional Matching of Extrema) to detect both clear and weak linear structures and to establish enough line-to-keypoint associations to represent this context. Evaluations on the TUM RGB-D SLAM dataset and outdoor video sequences show that CLAD provides geometrically consistent matches that complement those obtained with traditional local descriptors. In repetitive scenes, where local descriptors may not distinguish similar patches, CLAD uses global structure to improve matching reliability. The method is computationally lightweight and practical to implement.

Masaki Satoh

Morpho, Inc. Tokyo, Japan

Image Matching Local Feature Matching

Ask PDF

Oct 18, 2026

LateMatch: Breaking the Interaction Bottleneck via Cacheable Matching for Efficient Visual Localization

Visual localization is critical for embodied agents, but deploying state-of-the-art deep feature matchers on edge devices is fundamentally limited by their massive computational cost. Standard deep matchers tightly interleave intra-image context building and inter-image matching at every layer, creating a severe interaction bottleneck. Localizing a single query against K database images forces the system to execute the full network K independent times with no capacity for caching. To break this bottleneck, we advocate for a paradigm shift toward cacheable image matching. We realize this paradigm with LateMatch, a versatile approach that restructures existing matchers by deferring inter-image interaction to their final layers. By isolating intra-image processing into offline-cacheable Context Blocks, only a small set of Interaction Blocks must be executed online. To further optimize LateMatch, we introduce E-SToRE (Entropy-guided Spatial Token Reduction and Expansion). E-SToRE leverages the decreasing Shannon entropy of the late Interaction Blocks to dynamically reduce token counts just prior to cross-image interaction. Evaluated across HPatches, ScanNet, MegaDepth, ETH-3D, and Aachen, LateMatch reduces effective computational cost by over 2x compared to standard deep matchers, while E-SToRE yields an additional 10% to 14% savings with negligible degradation in accuracy.

Mohammad Omama, Aeyan Ashraf, Sandeep Chinchali

The University of Texas at Austin · University of Massachusetts Amherst

Image Matching Visual Place Recognition Edge Inference

Ask PDF

Aug 20, 2026

Unified and Efficient Point-Line Local Features

Multi-view computer vision pipelines typically rely on accurate sparse keypoints and robust descriptors. While incorporating line features has shown clear benefits for matching and pose estimation, existing point-line approaches remain inefficient: they detect points and lines separately, use increasingly heavy networks, and depend on CPU-bound heuristics that hinder real-time performance. We introduce a Unified Efficient Points and Lines (UPAL) feature extractor that jointly extracts keypoints, line segments, and feature descriptors within a single lightweight architecture. A shared backbone provides common representations that feed different branches for point and line features. Line segments are recovered through an accelerated post-processing stage, an enhanced and highly efficient variant of the LSD algorithm. UPAL matches or exceeds state-ofthe-art performance in both point and line applications while significantly reducing computational cost, achieving, for instance, a 4x speedup and 10x smaller memory footprint over the ALIKED + DeepLSD pipeline. Code is publicly available at https://github.com/francois141/upal.

François Costa, Raphael Kreft, Eckhard Goedeke +6

Department of Computer Science, ETH Zurich, Switzerland · Microsoft Spatial AI Lab

Local Feature Matching Keypoint Detection

Ask PDF

Abstract

Local Feature Matching

Explore similar work

CLAD: Keypoint Matching Using Global Line Correspondences

LateMatch: Breaking the Interaction Bottleneck via Cacheable Matching for Efficient Visual Localization

Unified and Efficient Point-Line Local Features