Jiřı́ Matas

Basit Alawode, Fayaz Ali Dharejo, Mehnaz Ummar et al.

This paper presents a new dataset and general tracker enhancement method for Underwater Visual Object Tracking (UVOT). Despite its significance, underwater tracking has remained unexplored due to data inaccessibility. It poses distinct challenges; the underwater environment exhibits non-uniform lighting conditions, low visibility, lack of sharpness, low contrast, camouflage, and reflections from suspended particles. Performance of traditional tracking methods designed primarily for terrestrial or open-air scenarios drops in such conditions. We address the problem by proposing a novel underwater image enhancement algorithm designed specifically to boost tracking quality. The method has resulted in a significant performance improvement, of up to 5.0% AUC, of state-of-the-art (SOTA) visual trackers. To develop robust and accurate UVOT methods, large-scale datasets are required. To this end, we introduce a large-scale UVOT benchmark dataset consisting of 400 video segments and 275,000 manually annotated frames enabling underwater training and evaluation of deep trackers. The videos are labelled with several underwater-specific tracking attributes including watercolor variation, target distractors, camouflage, target relative size, and low visibility conditions. The UVOT400 dataset, tracking results, and the code are publicly available on: https://github.com/BasitAlawode/UWVOT400.

Martin Sundermeyer, Tomas Hodan, Yann Labbe et al. · tsinghua

We present the evaluation methodology, datasets and results of the BOP Challenge 2022, the fourth in a series of public competitions organized with the goal to capture the status quo in the field of 6D object pose estimation from an RGB/RGB-D image. In 2022, we witnessed another significant improvement in the pose estimation accuracy -- the state of the art, which was 56.9 AR$_C$ in 2019 (Vidal et al.) and 69.8 AR$_C$ in 2020 (CosyPose), moved to new heights of 83.7 AR$_C$ (GDRNPP). Out of 49 pose estimation methods evaluated since 2019, the top 18 are from 2022. Methods based on point pair features, which were introduced in 2010 and achieved competitive results even in 2020, are now clearly outperformed by deep learning methods. The synthetic-to-real domain gap was again significantly reduced, with 82.7 AR$_C$ achieved by GDRNPP trained only on synthetic images from BlenderProc. The fastest variant of GDRNPP reached 80.5 AR$_C$ with an average time per image of 0.23s. Since most of the recent methods for 6D object pose estimation begin by detecting/segmenting objects, we also started evaluating 2D object detection and segmentation performance based on the COCO metrics. Compared to the Mask R-CNN results from CosyPose in 2020, detection improved from 60.3 to 77.3 AP$_C$ and segmentation from 40.5 to 58.7 AP$_C$. The online evaluation system stays open and is available at: \href{http://bop.felk.cvut.cz/}{bop.felk.cvut.cz}.

Štěpán Šimsa, Milan Šulc, Michal Uřičář et al.

This paper introduces the DocILE benchmark with the largest dataset of business documents for the tasks of Key Information Localization and Extraction and Line Item Recognition. It contains 6.7k annotated business documents, 100k synthetically generated documents, and nearly~1M unlabeled documents for unsupervised pre-training. The dataset has been built with knowledge of domain- and task-specific aspects, resulting in the following key features: (i) annotations in 55 classes, which surpasses the granularity of previously published key information extraction datasets by a large margin; (ii) Line Item Recognition represents a highly practical information extraction task, where key information has to be assigned to items in a table; (iii) documents come from numerous layouts and the test set includes zero- and few-shot cases as well as layouts commonly seen in the training set. The benchmark comes with several baselines, including RoBERTa, LayoutLMv3 and DETR-based Table Transformer; applied to both tasks of the DocILE benchmark, with results shared in this paper, offering a quick starting point for future work. The dataset, baselines and supplementary material are available at https://github.com/rossumai/docile.

Tong Wei, Yash Patel, Alexander Shekhovtsov et al.

We propose $\nabla$-RANSAC, a generalized differentiable RANSAC that allows learning the entire randomized robust estimation pipeline. The proposed approach enables the use of relaxation techniques for estimating the gradients in the sampling distribution, which are then propagated through a differentiable solver. The trainable quality function marginalizes over the scores from all the models estimated within $\nabla$-RANSAC to guide the network learning accurate and useful inlier probabilities or to train feature detection and matching networks. Our method directly maximizes the probability of drawing a good hypothesis, allowing us to learn better sampling distributions. We test $\nabla$-RANSAC on various real-world scenarios on fundamental and essential matrix estimation, and 3D point cloud registration, outdoors and indoors, with handcrafted and learning-based features. It is superior to the state-of-the-art in terms of accuracy while running at a similar speed to its less accurate alternatives. The code and trained models are available at https://github.com/weitong8591/differentiable_ransac.

Tomas Vojir, Jan Sochman, Rahaf Aljundi et al.

Out-of-distribution detection is a common issue in deploying vision models in practice and solving it is an essential building block in safety critical applications. Most of the existing OOD detection solutions focus on improving the OOD robustness of a classification model trained exclusively on in-distribution (ID) data. In this work, we take a different approach and propose to leverage generic pre-trained representation. We propose a novel OOD method, called GROOD, that formulates the OOD detection as a Neyman-Pearson task with well calibrated scores and which achieves excellent performance, predicated by the use of a good generic representation. Only a trivial training process is required for adapting GROOD to a particular problem. The method is simple, general, efficient, calibrated and with only a few hyper-parameters. The method achieves state-of-the-art performance on a number of OOD benchmarks, reaching near perfect performance on several of them. The source code is available at https://github.com/vojirt/GROOD.

Daniel Barath, Dmytro Mishkin, Michal Polic et al.

We present a large-scale dataset of Planes in 3D, Pi3D, of roughly 1000 planes observed in 10 000 images from the 1DSfM dataset, and HEB, a large-scale homography estimation benchmark leveraging Pi3D. The applications of the Pi3D dataset are diverse, e.g. training or evaluating monocular depth, surface normal estimation and image matching algorithms. The HEB dataset consists of 226 260 homographies and includes roughly 4M correspondences. The homographies link images that often undergo significant viewpoint and illumination changes. As applications of HEB, we perform a rigorous evaluation of a wide range of robust estimators and deep learning-based correspondence filtering methods, establishing the current state-of-the-art in robust homography estimation. We also evaluate the uncertainty of the SIFT orientations and scales w.r.t. the ground truth coming from the underlying homographies and provide codes for comparing uncertainty of custom detectors. The dataset is available at \url{https://github.com/danini/homography-benchmark}.

Jan Docekal, Jakub Rozlivek, Jiri Matas et al.

We study the performance of state-of-the-art human keypoint detectors in the context of close proximity human-robot interaction. The detection in this scenario is specific in that only a subset of body parts such as hands and torso are in the field of view. In particular, (i) we survey existing datasets with human pose annotation from the perspective of close proximity images and prepare and make publicly available a new Human in Close Proximity (HiCP) dataset; (ii) we quantitatively and qualitatively compare state-of-the-art human whole-body 2D keypoint detection methods (OpenPose, MMPose, AlphaPose, Detectron2) on this dataset; (iii) since accurate detection of hands and fingers is critical in applications with handovers, we evaluate the performance of the MediaPipe hand detector; (iv) we deploy the algorithms on a humanoid robot with an RGB-D camera on its head and evaluate the performance in 3D human keypoint detection. A motion capture system is used as reference. The best performing whole-body keypoint detectors in close proximity were MMPose and AlphaPose, but both had difficulty with finger detection. Thus, we propose a combination of MMPose or AlphaPose for the body and MediaPipe for the hands in a single framework providing the most accurate and robust detection. We also analyse the failure modes of individual detectors -- for example, to what extent the absence of the head of the person in the image degrades performance. Finally, we demonstrate the framework in a scenario where a humanoid robot interacting with a person uses the detected 3D keypoints for whole-body avoidance maneuvers.

Lukas Picek, Klara Janouskova, Vojtech Cermak et al.

We introduce a new, challenging benchmark and a dataset, FungiTastic, based on fungal records continuously collected over a twenty-year span. The dataset is labelled and curated by experts and consists of about 350k multimodal observations of 6k fine-grained categories (species). The fungi observations include photographs and additional data, e.g., meteorological and climatic data, satellite images, and body part segmentation masks. FungiTastic is one of the few benchmarks that include a test set with DNA-sequenced ground truth of unprecedented label reliability. The benchmark is designed to support (i) standard closed-set classification, (ii) open-set classification, (iii) multi-modal classification, (iv) few-shot learning, (v) domain shift, and many more. We provide tailored baselines for many use cases, a multitude of ready-to-use pre-trained models on https://huggingface.co/collections/BVRA/fungitastic-66a227ce0520be533dc6403b, and a framework for model training. The documentation and the baselines are available at https://github.com/BohemianVRA/FungiTastic/ and https://www.kaggle.com/datasets/picekl/fungitastic.

Jonas Serych, Jiri Matas

We propose WOFT -- a novel method for planar object tracking that estimates a full 8 degrees-of-freedom pose, i.e. the homography w.r.t. a reference view. The method uses a novel module that leverages dense optical flow and assigns a weight to each optical flow correspondence, estimating a homography by weighted least squares in a fully differentiable manner. The trained module assigns zero weights to incorrect correspondences (outliers) in most cases, making the method robust and eliminating the need of the typically used non-differentiable robust estimators like RANSAC. The proposed weighted optical flow tracker (WOFT) achieves state-of-the-art performance on two benchmarks, POT-210 and POIC, tracking consistently well across a wide range of scenarios.

Denys Rozumnyi, Jiri Matas, Marc Pollefeys et al.

Most model-free visual object tracking methods formulate the tracking task as object location estimation given by a 2D segmentation or a bounding box in each video frame. We argue that this representation is limited and instead propose to guide and improve 2D tracking with an explicit object representation, namely the textured 3D shape and 6DoF pose in each video frame. Our representation tackles a complex long-term dense correspondence problem between all 3D points on the object for all video frames, including frames where some points are invisible. To achieve that, the estimation is driven by re-rendering the input video frames as well as possible through differentiable rendering, which has not been used for tracking before. The proposed optimization minimizes a novel loss function to estimate the best 3D shape, texture, and 6DoF pose. We improve the state-of-the-art in 2D segmentation tracking on three different datasets with mostly rigid objects.

Basit Alawode, Sajid Javed, Arif Mahmood et al.

We observe that the performance of SOTA visual trackers surprisingly strongly varies across different video attributes and datasets. No single tracker remains the best performer across all tracking attributes and datasets. To bridge this gap, for a given video sequence, we predict the "Best of the N Trackers", called the BofN meta-tracker. At its core, a Tracking Performance Prediction Network (TP2N) selects a predicted best performing visual tracker for the given video sequence using only a few initial frames. We also introduce a frame-level BofN meta-tracker which keeps predicting best performer after regular temporal intervals. The TP2N is based on self-supervised learning architectures MocoV2, SwAv, BT, and DINO; experiments show that the DINO with ViT-S as a backbone performs the best. The video-level BofN meta-tracker outperforms, by a large margin, existing SOTA trackers on nine standard benchmarks - LaSOT, TrackingNet, GOT-10K, VOT2019, VOT2021, VOT2022, UAV123, OTB100, and WebUAV-3M. Further improvement is achieved by the frame-level BofN meta-tracker effectively handling variations in the tracking scenarios within long sequences. For instance, on GOT-10k, BofN meta-tracker average overlap is 88.7% and 91.1% with video and frame-level settings respectively. The best performing tracker, RTS, achieves 85.20% AO. On VOT2022, BofN expected average overlap is 67.88% and 70.98% with video and frame level settings, compared to the best performing ARTrack, 64.12%. This work also presents an extensive evaluation of competitive tracking methods on all commonly used benchmarks, following their protocols. The code, the trained models, and the results will soon be made publicly available on https://github.com/BasitAlawode/Best_of_N_Trackers.

Alan Lukezic, Ziga Trojer, Jiri Matas et al.

Visual object tracking has focused predominantly on opaque objects, while transparent object tracking received very little attention. Motivated by the uniqueness of transparent objects in that their appearance is directly affected by the background, the first dedicated evaluation dataset has emerged recently. We contribute to this effort by proposing the first transparent object tracking training dataset Trans2k that consists of over 2k sequences with 104,343 images overall, annotated by bounding boxes and segmentation masks. Noting that transparent objects can be realistically rendered by modern renderers, we quantify domain-specific attributes and render the dataset containing visual attributes and tracking situations not covered in the existing object training datasets. We observe a consistent performance boost (up to 16%) across a diverse set of modern tracking architectures when trained using Trans2k, and show insights not previously possible due to the lack of appropriate training sets. The dataset and the rendering engine will be publicly released to unlock the power of modern learning-based trackers and foster new designs in transparent object tracking.

Phong Nguyen-Ha, Lam Huynh, Esa Rahtu et al.

We present CG-NeRF, a cascade and generalizable neural radiance fields method for view synthesis. Recent generalizing view synthesis methods can render high-quality novel views using a set of nearby input views. However, the rendering speed is still slow due to the nature of uniformly-point sampling of neural radiance fields. Existing scene-specific methods can train and render novel views efficiently but can not generalize to unseen data. Our approach addresses the problems of fast and generalizing view synthesis by proposing two novel modules: a coarse radiance fields predictor and a convolutional-based neural renderer. This architecture infers consistent scene geometry based on the implicit neural fields and renders new views efficiently using a single GPU. We first train CG-NeRF on multiple 3D scenes of the DTU dataset, and the network can produce high-quality and accurate novel views on unseen real and synthetic data using only photometric losses. Moreover, our method can leverage a denser set of reference images of a single scene to produce accurate novel views without relying on additional explicit representations and still maintains the high-speed rendering of the pre-trained model. Experimental results show that CG-NeRF outperforms state-of-the-art generalizable neural rendering methods on various synthetic and real datasets.

Tetiana Martyniuk, Orest Kupyn, Yana Kurliak et al.

We present DAD-3DHeads, a dense and diverse large-scale dataset, and a robust model for 3D Dense Head Alignment in the wild. It contains annotations of over 3.5K landmarks that accurately represent 3D head shape compared to the ground-truth scans. The data-driven model, DAD-3DNet, trained on our dataset, learns shape, expression, and pose parameters, and performs 3D reconstruction of a FLAME mesh. The model also incorporates a landmark prediction branch to take advantage of rich supervision and co-training of multiple related tasks. Experimentally, DAD-3DNet outperforms or is comparable to the state-of-the-art models in (i) 3D Head Pose Estimation on AFLW2000-3D and BIWI, (ii) 3D Face Shape Reconstruction on NoW and Feng, and (iii) 3D Dense Head Alignment and 3D Landmarks Estimation on DAD-3DHeads dataset. Finally, the diversity of DAD-3DHeads in camera angles, facial expressions, and occlusions enables a benchmark to study in-the-wild generalization and robustness to distribution shifts. The dataset webpage is https://p.farm/research/dad-3dheads.

Lam Huynh, Esa Rahtu, Jiri Matas et al.

We present LDP, a lightweight dense prediction neural architecture search (NAS) framework. Starting from a pre-defined generic backbone, LDP applies the novel Assisted Tabu Search for efficient architecture exploration. LDP is fast and suitable for various dense estimation problems, unlike previous NAS methods that are either computational demanding or deployed only for a single subtask. The performance of LPD is evaluated on monocular depth estimation, semantic segmentation, and image super-resolution tasks on diverse datasets, including NYU-Depth-v2, KITTI, Cityscapes, COCO-stuff, DIV2K, Set5, Set14, BSD100, Urban100. Experiments show that the proposed framework yields consistent improvements on all tested dense prediction tasks, while being $5\%-315\%$ more compact in terms of the number of model parameters than prior arts.

Tuan-Hung Vu, Eduardo Valle, Andrei Bursuc et al.

We propose the unified BRAVO challenge to benchmark the reliability of semantic segmentation models under realistic perturbations and unknown out-of-distribution (OOD) scenarios. We define two categories of reliability: (1) semantic reliability, which reflects the model's accuracy and calibration when exposed to various perturbations; and (2) OOD reliability, which measures the model's ability to detect object classes that are unknown during training. The challenge attracted nearly 100 submissions from international teams representing notable research institutions. The results reveal interesting insights into the importance of large-scale pre-training and minimal architectural design in developing robust and reliable semantic segmentation models.

Vojtěch Cermak, Lukas Picek, Lukáš Adam et al.

We propose a new method - WildFusion - for individual identification of a broad range of animal species. The method fuses deep scores (e.g., MegaDescriptor or DINOv2) and local matching similarity (e.g., LoFTR and LightGlue) to identify individual animals. The global and local information fusion is facilitated by similarity score calibration. In a zero-shot setting, relying on local similarity score only, WildFusion achieved mean accuracy, measured on 17 datasets, of 76.2%. This is better than the state-of-the-art model, MegaDescriptor-L, whose training set included 15 of the 17 datasets. If a dataset-specific calibration is applied, mean accuracy increases by 2.3% percentage points. WildFusion, with both local and global similarity scores, outperforms the state-of-the-art significantly - mean accuracy reached 84.0%, an increase of 8.5 percentage points; the mean relative error drops by 35%. We make the code and pre-trained models publicly available5, enabling immediate use in ecology and conservation.

Lukas Picek, Lukas Neumann, Jiri Matas

We propose a method that robustly exploits background and foreground in visual identification of individual animals. Experiments show that their automatic separation, made easy with methods like Segment Anything, together with independent foreground and background-related modeling, improves results. The two predictions are combined in a principled way, thanks to novel Per-Instance Temperature Scaling that helps the classifier to deal with appearance ambiguities in training and to produce calibrated outputs in the inference phase. For identity prediction from the background, we propose novel spatial and temporal models. On two problems, the relative error w.r.t. the baseline was reduced by 22.3% and 8.8%, respectively. For cases where objects appear in new locations, an example of background drift, accuracy doubles.

Klara Janouskova, Tamir Shor, Chaim Baskin et al.

Test-Time Adaptation (TTA) methods improve the robustness of deep neural networks to domain shift on a variety of tasks such as image classification or segmentation. This work explores adapting segmentation models to a single unlabelled image with no other data available at test-time. In particular, this work focuses on adaptation by optimizing self-supervised losses at test-time. Multiple baselines based on different principles are evaluated under diverse conditions and a novel adversarial training is introduced for adaptation with mask refinement. Our additions to the baselines result in a 3.51 and 3.28 % increase over non-adapted baselines, without these improvements, the increase would be 1.7 and 2.16 % only.

Václav Vávra, Dmytro Mishkin, Jiří Matas

We consider the problem of two-view matching under significant viewpoint changes with view synthesis. We propose two novel methods, minimizing the view synthesis overhead. The first one, named DenseAffNet, uses dense affine shapes estimates from AffNet, which allows it to partition the image, rectifying each partition with just a single affine map. The second one, named DepthAffNet, combines information from depth maps and affine shapes estimates to produce different sets of rectifying affine maps for different image partitions. DenseAffNet is faster than the state-of-the-art and more accurate on generic scenes. DepthAffNet is on par with the state of the art on scenes containing large planes. The evaluation is performed on 3 public datasets - EVD Dataset, Strong ViewPoint Changes Dataset and IMC Phototourism Dataset.

Miroslav Purkrabek, Jiri Matas

Methods and datasets for human pose estimation focus predominantly on side- and front-view scenarios. We overcome the limitation by leveraging synthetic data and introduce RePoGen (RarE POses GENerator), an SMPL-based method for generating synthetic humans with comprehensive control over pose and view. Experiments on top-view datasets and a new dataset of real images with diverse poses show that adding the RePoGen data to the COCO dataset outperforms previous approaches to top- and bottom-view pose estimation without harming performance on common views. An ablation study shows that anatomical plausibility, a property prior research focused on, is not a prerequisite for effective performance. The introduced dataset and the corresponding code are available on https://mirapurkrabek.github.io/RePoGen-paper/ .

Andrii Yermakov, Jan Cech, Jiri Matas

This paper tackles the challenge of detecting partially manipulated facial deepfakes, which involve subtle alterations to specific facial features while retaining the overall context, posing a greater detection difficulty than fully synthetic faces. We leverage the Contrastive Language-Image Pre-training (CLIP) model, specifically its ViT-L/14 visual encoder, to develop a generalizable detection method that performs robustly across diverse datasets and unknown forgery techniques with minimal modifications to the original model. The proposed approach utilizes parameter-efficient fine-tuning (PEFT) techniques, such as LN-tuning, to adjust a small subset of the model's parameters, preserving CLIP's pre-trained knowledge and reducing overfitting. A tailored preprocessing pipeline optimizes the method for facial images, while regularization strategies, including L2 normalization and metric learning on a hyperspherical manifold, enhance generalization. Trained on the FaceForensics++ dataset and evaluated in a cross-dataset fashion on Celeb-DF-v2, DFDC, FFIW, and others, the proposed method achieves competitive detection accuracy comparable to or outperforming much more complex state-of-the-art techniques. This work highlights the efficacy of CLIP's visual encoder in facial deepfake detection and establishes a simple, powerful baseline for future research, advancing the field of generalizable deepfake detection. The code is available at: https://github.com/yermandy/deepfake-detection

Jonas Serych, Michal Neoral, Jiri Matas

In this work, we present MFTIQ, a novel dense long-term tracking model that advances the Multi-Flow Tracker (MFT) framework to address challenges in point-level visual tracking in video sequences. MFTIQ builds upon the flow-chaining concepts of MFT, integrating an Independent Quality (IQ) module that separates correspondence quality estimation from optical flow computations. This decoupling significantly enhances the accuracy and flexibility of the tracking process, allowing MFTIQ to maintain reliable trajectory predictions even in scenarios of prolonged occlusions and complex dynamics. Designed to be "plug-and-play", MFTIQ can be employed with any off-the-shelf optical flow method without the need for fine-tuning or architectural modifications. Experimental validations on the TAP-Vid Davis dataset show that MFTIQ with RoMa optical flow not only surpasses MFT but also performs comparably to state-of-the-art trackers while having substantially faster processing speed. Code and models available at https://github.com/serycjon/MFTIQ .

David Korcak, Jiri Matas

We present a method for estimating the shutter angle, a.k.a. exposure fraction - the ratio of the exposure time and the reciprocal of frame rate - of videoclips containing motion. The approach exploits the relation of the exposure fraction, optical flow, and linear motion blur. Robustness is achieved by selecting image patches where both the optical flow and blur estimates are reliable, checking their consistency. The method was evaluated on the publicly available Beam-Splitter Dataset with a range of exposure fractions from 0.015 to 0.36. The best achieved mean absolute error of estimates was 0.039. We successfully test the suitability of the method for a forensic application of detection of video tampering by frame removal or insertion

Andrii Yermakov, Jan Cech, Jiri Matas et al.

The generalization of deepfake detectors to unseen manipulation techniques remains a challenge for practical deployment. Although many approaches adapt foundation models by introducing significant architectural complexity, this work demonstrates that robust generalization is achievable through a parameter-efficient adaptation of one of the foundational pre-trained vision encoders. The proposed method, GenD, fine-tunes only the Layer Normalization parameters (0.03% of the total) and enhances generalization by enforcing a hyperspherical feature manifold using L2 normalization and metric learning on it. We conducted an extensive evaluation on 14 benchmark datasets spanning from 2019 to 2025. The proposed method achieves state-of-the-art performance, outperforming more complex, recent approaches in average cross-dataset AUROC. Our analysis yields two primary findings for the field: 1) training on paired real-fake data from the same source video is essential for mitigating shortcut learning and improving generalization, and 2) detection difficulty on academic datasets has not strictly increased over time, with models trained on older, diverse datasets showing strong generalization capabilities. This work delivers a computationally efficient and reproducible method, proving that state-of-the-art generalization is attainable by making targeted, minimal changes to a pre-trained foundational image encoder model. The code is at: https://github.com/yermandy/GenD

Tong Wei, Philipp Lindenberger, Jiri Matas et al.

Visual place recognition methods struggle with occlusions and partial visual overlaps. We propose a novel visual place recognition approach based on overlap prediction, called VOP, shifting from traditional reliance on global image similarities and local features to image overlap prediction. VOP proceeds co-visible image sections by obtaining patch-level embeddings using a Vision Transformer backbone and establishing patch-to-patch correspondences without requiring expensive feature detection and matching. Our approach uses a voting mechanism to assess overlap scores for potential database images. It provides a nuanced image retrieval metric in challenging scenarios. Experimental results show that VOP leads to more accurate relative pose estimation and localization results on the retrieved image pairs than state-of-the-art baselines on a number of large-scale, real-world indoor and outdoor benchmarks. The code is available at https://github.com/weitong8591/vop.git.

Vasyl Borsuk, Roman Vei, Orest Kupyn et al.

We present FEAR, a family of fast, efficient, accurate, and robust Siamese visual trackers. We present a novel and efficient way to benefit from dual-template representation for object model adaption, which incorporates temporal information with only a single learnable parameter. We further improve the tracker architecture with a pixel-wise fusion block. By plugging-in sophisticated backbones with the abovementioned modules, FEAR-M and FEAR-L trackers surpass most Siamese trackers on several academic benchmarks in both accuracy and efficiency. Employed with the lightweight backbone, the optimized version FEAR-XS offers more than 10 times faster tracking than current Siamese trackers while maintaining near state-of-the-art results. FEAR-XS tracker is 2.4x smaller and 4.3x faster than LightTrack with superior accuracy. In addition, we expand the definition of the model efficiency by introducing FEAR benchmark that assesses energy consumption and execution speed. We show that energy consumption is a limiting factor for trackers on mobile devices. Source code, pretrained models, and evaluation protocol are available at https://github.com/PinataFarms/FEARTracker.

Tong Wei, Jiri Matas, Daniel Barath

We propose a new sampler for robust estimators that always selects the sample with the highest probability of consisting only of inliers. After every unsuccessful iteration, the inlier probabilities are updated in a principled way via a Bayesian approach. The probabilities obtained by the deep network are used as prior (so-called neural guidance) inside the sampler. Moreover, we introduce a new loss that exploits, in a geometrically justifiable manner, the orientation and scale that can be estimated for any type of feature, e.g., SIFT or SuperPoint, to estimate two-view geometry. The new loss helps to learn higher-order information about the underlying scene geometry. Benefiting from the new sampler and the proposed loss, we combine the neural guidance with the state-of-the-art MAGSAC++. Adaptive Reordering Sampler with Neurally Guided MAGSAC (ARS-MAGSAC) is superior to the state-of-the-art in terms of accuracy and run-time on the PhotoTourism and KITTI datasets for essential and fundamental matrix estimation. The code and trained models are available at https://github.com/weitong8591/ars_magsac.

Daniel Barath, Denys Rozumny, Ivan Eichhardt et al.

We propose a new algorithm for finding an unknown number of geometric models, e.g., homographies. The problem is formalized as finding dominant model instances progressively without forming crisp point-to-model assignments. Dominant instances are found via a RANSAC-like sampling and a consolidation process driven by a model quality function considering previously proposed instances. New ones are found by clustering in the consensus space. This new formulation leads to a simple iterative algorithm with state-of-the-art accuracy while running in real-time on a number of vision problems - at least two orders of magnitude faster than the competitors on two-view motion estimation. Also, we propose a deterministic sampler reflecting the fact that real-world data tend to form spatially coherent structures. The sampler returns connected components in a progressively densified neighborhood-graph. We present a number of applications where the use of multiple geometric models improves accuracy. These include pose estimation from multiple generalized homographies; trajectory estimation of fast-moving objects; and we also propose a way of using multiple homographies in global SfM algorithms. Source code: https://github.com/danini/clustering-in-consensus-space.

Tomas Hodan, Martin Sundermeyer, Bertram Drost et al.

This paper presents the evaluation methodology, datasets, and results of the BOP Challenge 2020, the third in a series of public competitions organized with the goal to capture the status quo in the field of 6D object pose estimation from an RGB-D image. In 2020, to reduce the domain gap between synthetic training and real test RGB images, the participants were provided 350K photorealistic training images generated by BlenderProc4BOP, a new open-source and light-weight physically-based renderer (PBR) and procedural data generator. Methods based on deep neural networks have finally caught up with methods based on point pair features, which were dominating previous editions of the challenge. Although the top-performing methods rely on RGB-D image channels, strong results were achieved when only RGB channels were used at both training and test time - out of the 26 evaluated methods, the third method was trained on RGB channels of PBR and real images, while the fifth on RGB channels of PBR images only. Strong data augmentation was identified as a key component of the top-performing CosyPose method, and the photorealism of PBR images was demonstrated effective despite the augmentation. The online evaluation system stays open and is available on the project website: bop.felk.cvut.cz.

Daniel Barath, Jana Noskova, Jiri Matas

A method called, sigma-consensus, is proposed to eliminate the need for a user-defined inlier-outlier threshold in RANSAC. Instead of estimating the noise sigma, it is marginalized over a range of noise scales. The optimized model is obtained by weighted least-squares fitting where the weights come from the marginalization over sigma of the point likelihoods of being inliers. A new quality function is proposed not requiring sigma and, thus, a set of inliers to determine the model quality. Also, a new termination criterion for RANSAC is built on the proposed marginalization approach. Applying sigma-consensus, MAGSAC is proposed with no need for a user-defined sigma and improving the accuracy of robust estimation significantly. It is superior to the state-of-the-art in terms of geometric accuracy on publicly available real-world datasets for epipolar geometry (F and E) and homography estimation. In addition, applying sigma-consensus only once as a post-processing step to the RANSAC output always improved the model quality on a wide range of vision problems without noticeable deterioration in processing time, adding a few milliseconds. The source code is at https://github.com/danini/magsac.

Orest Kupyn, Volodymyr Budzan, Mykola Mykhailych et al.

We present DeblurGAN, an end-to-end learned method for motion deblurring. The learning is based on a conditional GAN and the content loss . DeblurGAN achieves state-of-the art performance both in the structural similarity measure and visual appearance. The quality of the deblurring model is also evaluated in a novel way on a real-world problem -- object detection on (de-)blurred images. The method is 5 times faster than the closest competitor -- DeepDeblur. We also introduce a novel method for generating synthetic motion blurred images from sharp ones, allowing realistic dataset augmentation. The model, code and the dataset are available at https://github.com/KupynOrest/DeblurGAN

Dmytro Mishkin, Filip Radenovic, Jiri Matas

A method for learning local affine-covariant regions is presented. We show that maximizing geometric repeatability does not lead to local regions, a.k.a features,that are reliably matched and this necessitates descriptor-based learning. We explore factors that influence such learning and registration: the loss function, descriptor type, geometric parametrization and the trade-off between matchability and geometric accuracy and propose a novel hard negative-constant loss function for learning of affine regions. The affine shape estimator -- AffNet -- trained with the hard negative-constant loss outperforms the state-of-the-art in bag-of-words image retrieval and wide baseline stereo. The proposed training process does not require precisely geometrically aligned patches.The source codes and trained weights are available at https://github.com/ducha-aiki/affnet

Giorgos Kordopatis-Zilos, Vladan Stojnić, Anna Manko et al.

This work introduces ILIAS, a new test dataset for Instance-Level Image retrieval At Scale. It is designed to evaluate the ability of current and future foundation models and retrieval techniques to recognize particular objects. The key benefits over existing datasets include large scale, domain diversity, accurate ground truth, and a performance that is far from saturated. ILIAS includes query and positive images for 1,000 object instances, manually collected to capture challenging conditions and diverse domains. Large-scale retrieval is conducted against 100 million distractor images from YFCC100M. To avoid false negatives without extra annotation effort, we include only query objects confirmed to have emerged after 2014, i.e. the compilation date of YFCC100M. An extensive benchmarking is performed with the following observations: i) models fine-tuned on specific domains, such as landmarks or products, excel in that domain but fail on ILIAS ii) learning a linear adaptation layer using multi-domain class supervision results in performance improvements, especially for vision-language models iii) local descriptors in retrieval re-ranking are still a key ingredient, especially in the presence of severe background clutter iv) the text-to-image performance of the vision-language foundation models is surprisingly close to the corresponding image-to-image case. website: https://vrg.fel.cvut.cz/ilias/

Nikita Kisel, Illia Volkov, Katerina Hanzelkova et al.

Since its release, ImageNet-1k dataset has become a gold standard for evaluating model performance. It has served as the foundation for numerous other datasets and training tasks in computer vision. As models have improved in accuracy, issues related to label correctness have become increasingly apparent. In this blog post, we analyze the issues in the ImageNet-1k dataset, including incorrect labels, overlapping or ambiguous class definitions, training-evaluation domain shifts, and image duplicates. The solutions for some problems are straightforward. For others, we hope to start a broader conversation about refining this influential dataset to better serve future research.

Jakub Kolář, Radim Špetlík, Jiří Matas

We introduce a novel method for measuring properties of periodic phenomena with an event camera, a device asynchronously reporting brightness changes at independently operating pixels. The approach assumes that for fast periodic phenomena, in any spatial window where it occurs, a very similar set of events is generated at the time difference corresponding to the frequency of the motion. To estimate the frequency, we compute correlations of spatio-temporal windows in the event space. The period is calculated from the time differences between the peaks of the correlation responses. The method is contactless, eliminating the need for markers, and does not need distinguishable landmarks. We evaluate the proposed method on three instances of periodic phenomena: (i) light flashes, (ii) vibration, and (iii) rotational speed. In all experiments, our method achieves a relative error lower than 0.04%, which is within the error margin of ground truth measurements.

Alan Lukezic, Ziga Trojer, Jiri Matas et al.

Performance of modern trackers degrades substantially on transparent objects compared to opaque objects. This is largely due to two distinct reasons. Transparent objects are unique in that their appearance is directly affected by the background. Furthermore, transparent object scenes often contain many visually similar objects (distractors), which often lead to tracking failure. However, development of modern tracking architectures requires large training sets, which do not exist in transparent object tracking. We present two contributions addressing the aforementioned issues. We propose the first transparent object tracking training dataset Trans2k that consists of over 2k sequences with 104,343 images overall, annotated by bounding boxes and segmentation masks. Standard trackers trained on this dataset consistently improve by up to 16%. Our second contribution is a new distractor-aware transparent object tracker (DiTra) that treats localization accuracy and target identification as separate tasks and implements them by a novel architecture. DiTra sets a new state-of-the-art in transparent object tracking and generalizes well to opaque objects.

Miroslav Purkrabek, Jiri Matas

Current Human Pose Estimation methods have achieved significant improvements. However, state-of-the-art models ignore out-of-image keypoints and use uncalibrated heatmaps as keypoint location representations. To address these limitations, we propose ProbPose, which predicts for each keypoint: a calibrated probability of keypoint presence at each location in the activation window, the probability of being outside of it, and its predicted visibility. To address the lack of evaluation protocols for out-of-image keypoints, we introduce the CropCOCO dataset and the Extended OKS (Ex-OKS) metric, which extends OKS to out-of-image points. Tested on COCO, CropCOCO, and OCHuman, ProbPose shows significant gains in out-of-image keypoint localization while also improving in-image localization through data augmentation. Additionally, the model improves robustness along the edges of the bounding box and offers better flexibility in keypoint evaluation. The code and models are available on https://mirapurkrabek.github.io/ProbPose/ for research purposes.

Tomáš Jelínek, Jonáš Šerých, Jiří Matas

Optical flow is a useful input for various applications, including 3D reconstruction, pose estimation, tracking, and structure-from-motion. Despite its utility, the field of dense long-term tracking, especially over wide baselines, has not been extensively explored. This paper extends the concept of combining multiple optical flows over logarithmically spaced intervals as proposed by MFT. We demonstrate the compatibility of MFT with different optical flow networks, yielding results that surpass their individual performance. Moreover, we present a simple yet effective combination of these networks within the MFT framework. This approach proves to be competitive with more sophisticated, non-causal methods in terms of position prediction accuracy, highlighting the potential of MFT in enhancing long-term tracking applications.

Miroslav Purkrabek, Jiri Matas

Human pose estimation methods work well on isolated people but struggle with multiple-bodies-in-proximity scenarios. Previous work has addressed this problem by conditioning pose estimation by detected bounding boxes or keypoints, but overlooked instance masks. We propose to iteratively enforce mutual consistency of bounding boxes, instance masks, and poses. The introduced BBox-Mask-Pose (BMP) method uses three specialized models that improve each other's output in a closed loop. All models are adapted for mutual conditioning, which improves robustness in multi-body scenes. MaskPose, a new mask-conditioned pose estimation model, is the best among top-down approaches on OCHuman. BBox-Mask-Pose pushes SOTA on OCHuman dataset in all three tasks - detection, instance segmentation, and pose estimation. It also achieves SOTA performance on COCO pose estimation. The method is especially good in scenes with large instances overlap, where it improves detection by 39% over the baseline detector. With small specialized models and faster runtime, BMP is an effective alternative to large human-centered foundational models. Code and models are available on https://MiraPurkrabek.github.io/BBox-Mask-Pose.

Matej Suchanek, Klara Janouskova, Ondrej Vasatko et al.

Visual-language models such as CLIP provide powerful general-purpose representations, but their raw embeddings are not optimized for supervised classification, often exhibiting limited class separation and excessive dimensionality. We propose Koo-Fu CLIP, a supervised CLIP adaptation method based on Fukunaga-Koontz Linear Discriminant Analysis, which operates in a whitened embedding space to suppress within-class variation and enhance between-class discrimination. The resulting closed-form linear projection reshapes the geometry of CLIP embeddings, improving class separability while performing effective dimensionality reduction, and provides a lightweight and efficient adaptation of CLIP representations. Across large-scale ImageNet benchmarks, nearest visual prototype classification in the Koo-Fu CLIP space improves top-1 accuracy from 75.1% to 79.1% on ImageNet-1K, with consistent gains persisting as the label space expands to 14K and 21K classes. The method supports substantial compression by up to 10-12x with little or no loss in accuracy, enabling efficient large-scale classification and retrieval.

Constantin Kolomiiets, Miroslav Purkrabek, Jiri Matas

Segment Anything (SAM) provides an unprecedented foundation for human segmentation, but may struggle under occlusion, where keypoints may be partially or fully invisible. We adapt SAM 2.1 for pose-guided segmentation with minimal encoder modifications, retaining its strong generalization. Using a fine-tuning strategy called PoseMaskRefine, we incorporate pose keypoints with high visibility into the iterative correction process originally employed by SAM, yielding improved robustness and accuracy across multiple datasets. During inference, we simplify prompting by selecting only the three keypoints with the highest visibility. This strategy reduces sensitivity to common errors, such as missing body parts or misclassified clothing, and allows accurate mask prediction from as few as a single keypoint. Our results demonstrate that pose-guided fine-tuning of SAM enables effective, occlusion-aware human segmentation while preserving the generalization capabilities of the original model. The code and pretrained models will be available at https://mirapurkrabek.github.io/BBox-MaskPose.

Matej Suchanek, Miroslav Purkrabek, Jiri Matas

UV map estimation is used in computer vision for detailed analysis of human posture or activity. Previous methods assign pixels to body model vertices by comparing pixel descriptors independently, without enforcing global coherence or plausibility in the UV map. We propose Pose-Constrained Continuous Surface Embeddings (PC-CSE), which integrates estimated 2D human pose into the pixel-to-vertex assignment process. The pose provides global anatomical constraints, ensuring that UV maps remain coherent while preserving local precision. Evaluation on DensePose COCO demonstrates consistent improvement, regardless of the chosen 2D human pose model. Whole-body poses offer better constraints by incorporating additional details about the hands and feet. Conditioning UV maps with human pose reduces invalid mappings and enhances anatomical plausibility. In addition, we highlight inconsistencies in the ground-truth annotations.

Yash Patel, Giorgos Tolias, Jiri Matas

This paper addresses supervised deep metric learning for open-set image retrieval, focusing on three key aspects: the loss function, mixup regularization, and model initialization. In deep metric learning, optimizing the retrieval evaluation metric, recall@k, via gradient descent is desirable but challenging due to its non-differentiable nature. To overcome this, we propose a differentiable surrogate loss that is computed on large batches, nearly equivalent to the entire training set. This computationally intensive process is made feasible through an implementation that bypasses the GPU memory limitations. Additionally, we introduce an efficient mixup regularization technique that operates on pairwise scalar similarities, effectively increasing the batch size even further. The training process is further enhanced by initializing the vision encoder using foundational models, which are pre-trained on large-scale datasets. Through a systematic study of these components, we demonstrate that their synergy enables large models to nearly solve popular benchmarks.

Klara Janouskova, Cristian Gavrus, Jiri Matas

In object recognition, both the subject of interest (referred to as foreground, FG, for simplicity) and its surrounding context (background, BG) may play an important role. However, standard supervised learning often leads to unintended over-reliance on the BG, limiting model robustness in real-world deployment settings. The problem is mainly addressed by suppressing the BG, sacrificing context information for improved generalization. We propose "Localize to Recognize Robustly" (L2R2), a novel recognition approach which exploits the benefits of context-aware classification while maintaining robustness to distribution shifts. L2R2 leverages advances in zero-shot detection to localize the FG before recognition. It improves the performance of both standard recognition with supervised training, as well as multimodal zero-shot recognition with VLMs, while being robust to long-tail BGs and distribution shifts. The results confirm localization before recognition is possible for a wide range of datasets and they highlight the limits of object detection on others

Tomas Hodan, Martin Sundermeyer, Yann Labbe et al.

We present the evaluation methodology, datasets and results of the BOP Challenge 2023, the fifth in a series of public competitions organized to capture the state of the art in model-based 6D object pose estimation from an RGB/RGB-D image and related tasks. Besides the three tasks from 2022 (model-based 2D detection, 2D segmentation, and 6D localization of objects seen during training), the 2023 challenge introduced new variants of these tasks focused on objects unseen during training. In the new tasks, methods were required to learn new objects during a short onboarding stage (max 5 minutes, 1 GPU) from provided 3D object models. The best 2023 method for 6D localization of unseen objects (GenFlow) notably reached the accuracy of the best 2020 method for seen objects (CosyPose), although being noticeably slower. The best 2023 method for seen objects (GPose) achieved a moderate accuracy improvement but a significant 43% run-time improvement compared to the best 2022 counterpart (GDRNPP). Since 2017, the accuracy of 6D localization of seen objects has improved by more than 50% (from 56.9 to 85.6 AR_C). The online evaluation system stays open and is available at: http://bop.felk.cvut.cz/.

Michal Neoral, Jonáš Šerých, Jiří Matas

We propose MFT -- Multi-Flow dense Tracker -- a novel method for dense, pixel-level, long-term tracking. The approach exploits optical flows estimated not only between consecutive frames, but also for pairs of frames at logarithmically spaced intervals. It selects the most reliable sequence of flows on the basis of estimates of its geometric accuracy and the probability of occlusion, both provided by a pre-trained CNN. We show that MFT achieves competitive performance on the TAP-Vid benchmark, outperforming baselines by a significant margin, and tracking densely orders of magnitude faster than the state-of-the-art point-tracking methods. The method is insensitive to medium-length occlusions and it is robustified by estimating flow with respect to the reference frame, which reduces drift.

Alan Lukežič, Jiří Matas, Matej Kristan

Template-based discriminative trackers are currently the dominant tracking paradigm due to their robustness, but are restricted to bounding box tracking and a limited range of transformation models, which reduces their localization accuracy. We propose a discriminative single-shot segmentation tracker -- D3S2, which narrows the gap between visual object tracking and video object segmentation. A single-shot network applies two target models with complementary geometric properties, one invariant to a broad range of transformations, including non-rigid deformations, the other assuming a rigid object to simultaneously achieve robust online target segmentation. The overall tracking reliability is further increased by decoupling the object and feature scale estimation. Without per-dataset finetuning, and trained only for segmentation as the primary output, D3S2 outperforms all published trackers on the recent short-term tracking benchmark VOT2020 and performs very close to the state-of-the-art trackers on the GOT-10k, TrackingNet, OTB100 and LaSoT. D3S2 outperforms the leading segmentation tracker SiamMask on video object segmentation benchmarks and performs on par with top video object segmentation algorithms.

Sajid Javed, Martin Danelljan, Fahad Shahbaz Khan et al.

Accurate and robust visual object tracking is one of the most challenging and fundamental computer vision problems. It entails estimating the trajectory of the target in an image sequence, given only its initial location, and segmentation, or its rough approximation in the form of a bounding box. Discriminative Correlation Filters (DCFs) and deep Siamese Networks (SNs) have emerged as dominating tracking paradigms, which have led to significant progress. Following the rapid evolution of visual object tracking in the last decade, this survey presents a systematic and thorough review of more than 90 DCFs and Siamese trackers, based on results in nine tracking benchmarks. First, we present the background theory of both the DCF and Siamese tracking core formulations. Then, we distinguish and comprehensively review the shared as well as specific open research challenges in both these tracking paradigms. Furthermore, we thoroughly analyze the performance of DCF and Siamese trackers on nine benchmarks, covering different experimental aspects of visual tracking: datasets, evaluation metrics, performance, and speed comparisons. We finish the survey by presenting recommendations and suggestions for distinguished open challenges based on our analysis.

Xiaoyan Xing, Yanlin Qian, Sibo Feng et al.

In this paper, we present Point Cloud Color Constancy, in short PCCC, an illumination chromaticity estimation algorithm exploiting a point cloud. We leverage the depth information captured by the time-of-flight (ToF) sensor mounted rigidly with the RGB sensor, and form a 6D cloud where each point contains the coordinates and RGB intensities, noted as (x,y,z,r,g,b). PCCC applies the PointNet architecture to the color constancy problem, deriving the illumination vector point-wise and then making a global decision about the global illumination chromaticity. On two popular RGB-D datasets, which we extend with illumination information, as well as on a novel benchmark, PCCC obtains lower error than the state-of-the-art algorithms. Our method is simple and fast, requiring merely 16*16-size input and reaching speed over 500 fps, including the cost of building the point cloud and net inference.