Skip to content

๐ŸŽฏ Object Detection

๐Ÿง  NeurIPS2026 ยท 17 paper notes

๐Ÿ“Œ Same area in other venues: ๐ŸŽž๏ธ ECCV2026 (78) ยท ๐Ÿ“ท CVPR2026 (99) ยท ๐Ÿ”ฌ ICLR2026 (30) ยท ๐Ÿงช ICML2026 (7) ยท ๐Ÿค– AAAI2026 (29) ยท ๐Ÿง  NeurIPS2025 (27)

๐Ÿ”ฅ Top topics: Anomaly Detection ร—6 ยท Object Detection ร—6 ยท Few-/Zero-Shot Learning ร—3 ยท Domain Adaptation ร—2

Align3D-AD: Cross-Modal Feature Alignment and Dual-Prompt Learning for Zero-Shot 3D Anomaly Detection

Addressing the severe domain gap between geometric 3D renderings and RGB-pretrained vision encoders, Align3D-AD leverages auxiliary RGB supervision during training via feature alignment with semantic consistency reweighting and couples it with modality-aware dual-prompt contrastive learning, achieving state-of-the-art zero-shot 3D anomaly detection using purely 3D inputs during inference.

Beyond Normal References: Discriminative Few-Shot Anomaly Detection

IDEAL uses a fixed set of few normal and masked anomalous references to suppress local normal variations, extract diverse intrinsic deviation directions, and score their projections, improving seen and unseen anomaly detection without target-domain retraining; under N1A1, it achieves 96.3 / 96.8 image-level / pixel-level AUROC on MVTecAD.

BiHashFormer: Hash-Driven Dual-Branch Transformer for Efficient Object Detection in HRW Shots

To overcome extreme foreground sparsity, computational explosion, and heavy background interleaving in gigapixel-level High-Resolution Wide (HRW) shots, BiHashFormer introduces a hash-driven dual-branch vision transformer that discards half of the key-value pairs in selected RoIs via HashAttention and recovers overlooked targets via HashMiner in non-RoI regions, reducing backbone FLOPs by 37.5% while boosting detection accuracy on the PANDA benchmark.

Contrastive Hypergraph Source-free Domain Adaptive Object Detection in Adverse Weather Conditions

Addressing source-free domain adaptive object detection for autonomous driving under adverse conditions like fog, rain, snow, and night, REUNION formulates high-order contextual modeling via multi-relational hypergraphs and optimizes a group-wise Hyper-InfoNCE objective to propagate semantics from confident instances to corrupted proposals, establishing new state-of-the-art detection accuracy on Cityscapes-Adverse.

COVD: Continual Open-Vocabulary Object Detection with Novel Concept Injection

Addressing the challenge of continually emerging concepts in real-world scenarios without expensive retraining, this paper proposes the Continual OVD with Novel Concept Injection (COVD) task alongside the Novel-114 benchmark, and designs NoIn-Detโ€”a parameter-free expansion framework that freezes the visual encoder and updates only text-branch parameters via stability distillation and knowledge-aware decoupling to assimilate novel concepts while preserving base open-vocabulary capabilities.

Detect Anything in Graphic Design: Element-Level Rewards for Autoregressive Detection

Addressing layer overlap and heavy occlusion in graphic designs, this paper presents DAD, an autoregressive amodal detection model framed as compositional deconstruction, combined with Element Relative Policy Optimization (EleRPO) based on martingale difference decomposition to achieve human-level occluded region recovery and layer decomposition.

Distilling Multi-Teacher Scoring Principles for Unsupervised Time Series Anomaly Detection

Addressing the inference overhead of multi-detector ensembles in unsupervised time-series anomaly detection, AnoMix introduces the CoRank objective to convert normalized multi-teacher score gaps into pairwise preference and consensus confidence supervision, training a compact single-head student (<80k parameters) to directly output scalar anomaly scores that outperform individual detectors and static ensembles across seven benchmarks, with strong zero-shot cross-dataset transferability.

Dynamic Query as Budgeting for Tiny Object Detection

This paper demonstrates that dynamic query mechanisms in tiny object detection primarily function as decoder capacity budgeting rather than fine-grained query selection, and proposes BUTR, an ultra-lightweight, non-learned budgeting-based DETR that outperforms prior complex selection designs with near-zero overhead.

GIFT: Representation Geometry Matters for Single-Domain Generalized Object Detection

Addressing the degradation caused by unconstrained fine-tuning that disrupts pre-trained representation geometry in Vision Foundation Models under single-source supervision, this paper proposes Geometric-Invariance Fine-Tuning (GIFT), which preserves relational geometry via orthogonal Householder reflections and adapts task-specific features via channel-wise scaling, outperforming prior SOTA by +7.9% and +2.9% mPC on Cityscapes-C and DWD with fewer than 1% trainable parameters.

Localize Any Object in X-Ray Security Scans without Human Annotation

To tackle volumetric superposition, weak boundaries, and the severe scarcity of dense annotations in X-ray security inspection, this paper proposes LAO-X, a self-supervised adaptation framework that mines isolated objects from unlabeled scans, synthesizes realistic overlapping scans via the physical Beerโ€“Lambert law, and fine-tunes SAM2 under an occlusion-controlled curriculum with local-to-holistic multi-granularity supervision.

M\(^2\)E-UAV: A Benchmark and Analysis for Onboard Motion-on-Motion Event-Based Tiny UAV Detection

This paper introduces M\(^2\)E-UAV, the first benchmark for onboard motion-on-motion event-based tiny UAV detection, providing synchronized 720p event streams, IMU data, and event-level foreground labels to reveal how ego-motion-induced background clutter severely degrades existing detectors.

Neurosymbolic Object-Centric Learning with Distant Supervision

DeepObjectLog integrates a slot-based perceptual encoder with a ProbLog probabilistic logic layer, enabling end-to-end learning of object-centric representations from distant, task-level labels alone via exact marginalization over latent objectness and class variables, without requiring bounding boxes, masks, or set-matching losses.

One Language-Free Foundation Model Is Enough for Universal Vision Anomaly Detection

Uncovering that the text encoder in vision-language anomaly detection fundamentally serves only to derive decision weights for anomaly classification and segmentation, this paper introduces UniADet, an embarrassingly simple language-free universal anomaly detection framework that completely decouples task weights and multi-scale hierarchical feature weights, achieving state-of-the-art performance across 14 industrial and medical benchmarks with merely 0.002M learnable parameters and outperforming full-shot methods in few-shot settings.

Order Matters: Competition-Guided Query Ordering for RNN-Based Object Detection

Addressing the issue where parallel order-agnostic self-attention in DETR decoders fails to organize query competition and leaves severe prediction redundancy in crowded scenes, DETRNN constructs an explicit competition-aware query sequence based on confidence and similarity, then applies a bidirectional RNN (such as Mamba) for recurrent refinement to softly suppress duplicate hypotheses without discarding queries.

Rethinking Infrared Small Target Detection: A Foundation-Driven Efficient Paradigm

To resolve the dilemma between global context deficiency and severe inference overhead in single-frame infrared small target (SIRST) detection, this paper introduces FDEP, a foundation-driven efficient paradigm that transfers frozen visual foundation model (VFM) priors via semantic alignment modulation fusion (SAMF) and parameter-shared implicit self-distillation (CO-ISD) with zero inference overhead, coupled with a multi-threshold integral evaluation metric (HSE).

TokenCLIP: Token-wise Prompt Learning for Zero-shot Anomaly Detection

TokenCLIP overcomes the representation bottleneck of indiscriminate text-image alignment in zero-shot anomaly detection by projecting the textual space into orthogonal subspaces and dynamically assigning them to visual patch tokens via Optimal Transport, achieving token-level customized supervision with minimal computational overhead.

Towards Explainable Industrial Anomaly Detection via Knowledge-Guided Latent Reasoning

To address the lack of domain-specific defect knowledge and the inefficiency of explicit reasoning chains in general-domain MLLMs, this paper proposes Reason-IADโ€”the first training-free test-time latent reasoning framework that incorporates retrieval-augmented priors and iteratively optimizes latent think tokens and dynamic visual evidence injection via entropy-based policy gradients.