In the past ten years, with the help of deep learning, especially the rapid development of deep neural networks, medical image analysis has made remarkable progress. However, how to effectively use the relational information between various tissues or organs in medical images is still a very challenging problem, and it has not been fully studied. In this thesis, we propose two novel solutions to this problem based on deep relational learning. First, we propose a context-aware fully convolutional network that effectively models implicit relation information between features to perform medical image segmentation. The network achieves the state-of-the-art segmentation results on the Multi Modal Brain Tumor Segmentation 2017 (BraTS2017) and Multi Modal Brain Tumor Segmentation 2018 (BraTS2018) data sets. Subsequently, we propose a new hierarchical homography estimation network to achieve accurate medical image mosaicing by learning the explicit spatial relationship between adjacent frames. We use the UCL Fetoscopy Placenta dataset to conduct experiments and our hierarchical homography estimation network outperforms the other state-of-the-art mosaicing methods while generating robust and meaningful mosaicing result on unseen frames.
The Lucas-Kanade (LK) method is a classic iterative homography estimation algorithm for image alignment, but often suffers from poor local optimality especially when image pairs have large distortions. To address this challenge, in this paper we propose a novel Deep Star-Convexified Lucas-Kanade (PRISE) method for multimodel image alignment by introducing strongly star-convex constraints into the optimization problem. Our basic idea is to enforce the neural network to approximately learn a star-convex loss landscape around the ground truth give any data to facilitate the convergence of the LK method to the ground truth through the high dimensional space defined by the network. This leads to a minimax learning problem, with contrastive (hinge) losses due to the definition of strong star-convexity that are appended to the original loss for training. We also provide an efficient sampling based algorithm to leverage the training cost, as well as some analysis on the quality of the solutions from PRISE. We further evaluate our approach on benchmark datasets such as MSCOCO, GoogleEarth, and GoogleMap, and demonstrate state-of-the-art results, especially for small pixel errors. Code can be downloaded from https://github.com/Zhang-VISLab.
Feature-based homography estimation approaches rely on extensive image processing for feature extraction and matching, and do not adequately account for the information provided by the image. Therefore, developing efficient direct techniques to extract the homography from images is essential. This paper presents a novel nonlinear direct homography observer that exploits the Lie group structure of $\mathbf{SL}(3)$ and its action on the space of image maps. Theoretical analysis demonstrates local asymptotic convergence of the observer. The observer design is also extended for partial measurements of velocity under the assumption that the unknown component is constant or slowly time-varying. Finally, simulation results demonstrate the performance of the proposed solutions on real images.
Heavy computation is a bottleneck limiting deep-learningbased feature matching algorithms to be applied in many realtime applications. However, existing lightweight networks optimized for Euclidean data cannot address classical feature matching tasks, since sparse keypoint based descriptors are expected to be matched. This paper tackles this problem and proposes two concepts: 1) a novel parallel attention model entitled ParaFormer and 2) a graph based U-Net architecture with attentional pooling. First, ParaFormer fuses features and keypoint positions through the concept of amplitude and phase, and integrates self- and cross-attention in a parallel manner which achieves a win-win performance in terms of accuracy and efficiency. Second, with U-Net architecture and proposed attentional pooling, the ParaFormer-U variant significantly reduces computational complexity, and minimize performance loss caused by downsampling. Sufficient experiments on various applications, including homography estimation, pose estimation, and image matching, demonstrate that ParaFormer achieves state-of-the-art performance while maintaining high efficiency. The efficient ParaFormer-U variant achieves comparable performance with less than 50% FLOPs of the existing attention-based models.
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}.
Existing homography and optical flow methods are erroneous in challenging scenes, such as fog, rain, night, and snow because the basic assumptions such as brightness and gradient constancy are broken. To address this issue, we present an unsupervised learning approach that fuses gyroscope into homography and optical flow learning. Specifically, we first convert gyroscope readings into motion fields named gyro field. Second, we design a self-guided fusion module (SGF) to fuse the background motion extracted from the gyro field with the optical flow and guide the network to focus on motion details. Meanwhile, we propose a homography decoder module (HD) to combine gyro field and intermediate results of SGF to produce the homography. To the best of our knowledge, this is the first deep learning framework that fuses gyroscope data and image content for both deep homography and optical flow learning. To validate our method, we propose a new dataset that covers regular and challenging scenes. Experiments show that our method outperforms the state-of-the-art methods in both regular and challenging scenes.
Homography estimation is a basic computer vision task, which aims to obtain the transformation from multi-view images for image alignment. Unsupervised learning homography estimation trains a convolution neural network for feature extraction and transformation matrix regression. While the state-of-theart homography method is based on convolution neural networks, few work focuses on transformer which shows superiority in highlevel vision tasks. In this paper, we propose a strong-baseline model based on the Swin Transformer, which combines convolution neural network for local features and transformer module for global features. Moreover, a cross non-local layer is introduced to search the matched features within the feature maps coarsely. In the homography regression stage, we adopt an attention layer for the channels of correlation volume, which can drop out some weak correlation feature points. The experiment shows that in 8 Degree-of-Freedoms(DOFs) homography estimation our method overperforms the state-of-the-art method.
Homography estimation is erroneous in the case of large-baseline due to the low image overlay and limited receptive field. To address it, we propose a progressive estimation strategy by converting large-baseline homography into multiple intermediate ones, cumulatively multiplying these intermediate items can reconstruct the initial homography. Meanwhile, a semi-supervised homography identity loss, which consists of two components: a supervised objective and an unsupervised objective, is introduced. The first supervised loss is acting to optimize intermediate homographies, while the second unsupervised one helps to estimate a large-baseline homography without photometric losses. To validate our method, we propose a large-scale dataset that covers regular and challenging scenes. Experiments show that our method achieves state-of-the-art performance in large-baseline scenes while keeping competitive performance in small-baseline scenes. Code and dataset are available at https://github.com/megvii-research/LBHomo.
The traditional homography estimation pipeline consists of four main steps: feature detection, feature matching, outlier removal and transformation estimation. Recent deep learning models intend to address the homography estimation problem using a single convolutional network. While these models are trained in an end-to-end fashion to simplify the homography estimation problem, they lack the feature matching step and/or the outlier removal step, which are important steps in the traditional homography estimation pipeline. In this paper, we attempt to build a deep learning model that mimics all four steps in the traditional homography estimation pipeline. In particular, the feature matching step is implemented using the cost volume technique. To remove outliers in the cost volume, we treat this outlier removal problem as a denoising problem and propose a novel self-supervised loss to solve the problem. Extensive experiments on synthetic and real datasets demonstrate that the proposed model outperforms existing deep learning models.
Generating robust and reliable correspondences across images is a fundamental task for a diversity of applications. To capture context at both global and local granularity, we propose ASpanFormer, a Transformer-based detector-free matcher that is built on hierarchical attention structure, adopting a novel attention operation which is capable of adjusting attention span in a self-adaptive manner. To achieve this goal, first, flow maps are regressed in each cross attention phase to locate the center of search region. Next, a sampling grid is generated around the center, whose size, instead of being empirically configured as fixed, is adaptively computed from a pixel uncertainty estimated along with the flow map. Finally, attention is computed across two images within derived regions, referred to as attention span. By these means, we are able to not only maintain long-range dependencies, but also enable fine-grained attention among pixels of high relevance that compensates essential locality and piece-wise smoothness in matching tasks. State-of-the-art accuracy on a wide range of evaluation benchmarks validates the strong matching capability of our method.
The Global Positioning System (GPS) has become a part of our daily life with the primary goal of providing geopositioning service. For an unmanned aerial system (UAS), geolocalization ability is an extremely important necessity which is achieved using Inertial Navigation System (INS) with the GPS at its heart. Without geopositioning service, UAS is unable to fly to its destination or come back home. Unfortunately, GPS signals can be jammed and suffer from a multipath problem in urban canyons. Our goal is to propose an alternative approach to geolocalize a UAS when GPS signal is degraded or denied. Considering UAS has a downward-looking camera on its platform that can acquire real-time images as the platform flies, we apply modern deep learning techniques to achieve geolocalization. In particular, we perform image matching to establish latent feature conjugates between UAS acquired imagery and satellite orthophotos. A typical application of feature matching suffers from high-rise buildings and new constructions in the field that introduce uncertainties into homography estimation, hence results in poor geolocalization performance. Instead, we extract GIS information from OpenStreetMap (OSM) to semantically segment matched features into building and terrain classes. The GIS mask works as a filter in selecting semantically matched features that enhance coplanarity conditions and the UAS geolocalization accuracy. Once the paper is published our code will be publicly available at https://github.com/OSUPCVLab/UbihereDrone2021.
Sports field registration in broadcast videos is typically interpreted as the task of homography estimation, which provides a mapping between a planar field and the corresponding visible area of the image. In contrast to previous approaches, we consider the task as a camera calibration problem. First, we introduce a differentiable objective function that is able to learn the camera pose and focal length from segment correspondences (e.g., lines, point clouds), based on pixel-level annotations for segments of a known calibration object. The calibration module iteratively minimizes the segment reprojection error induced by the estimated camera parameters. Second, we propose a novel approach for 3D sports field registration from broadcast soccer images. Compared to the typical solution, which subsequently refines an initial estimation, our solution does it in one step. The proposed method is evaluated for sports field registration on two datasets and achieves superior results compared to two state-of-the-art approaches.
We tackle the essential task of finding dense visual correspondences between a pair of images. This is a challenging problem due to various factors such as poor texture, repetitive patterns, illumination variation, and motion blur in practical scenarios. In contrast to methods that use dense correspondence ground-truths as direct supervision for local feature matching training, we train 3DG-STFM: a multi-modal matching model (Teacher) to enforce the depth consistency under 3D dense correspondence supervision and transfer the knowledge to 2D unimodal matching model (Student). Both teacher and student models consist of two transformer-based matching modules that obtain dense correspondences in a coarse-to-fine manner. The teacher model guides the student model to learn RGB-induced depth information for the matching purpose on both coarse and fine branches. We also evaluate 3DG-STFM on a model compression task. To the best of our knowledge, 3DG-STFM is the first student-teacher learning method for the local feature matching task. The experiments show that our method outperforms state-of-the-art methods on indoor and outdoor camera pose estimations, and homography estimation problems. Code is available at: https://github.com/Ryan-prime/3DG-STFM.
Homography has an essential relationship with the special linear group and the embedding Lie algebra structure. Although the Lie algebra representation is elegant, few researchers have established the connection between homography and algebra expression in neural networks. In this paper, we propose Warped Convolution Networks (WCN) to effectively learn and represent the homography by SL(3) group and sl(3) algebra with group convolution. To this end, six commutative subgroups within the SL(3) group are composed to form a homography. For each subgroup, a warping function is proposed to bridge the Lie algebra structure to its corresponding parameters in homography. By taking advantage of the warped convolution, homography learning is formulated into several simple pseudo-translation regressions. By walking along the Lie topology, our proposed WCN is able to learn the features that are invariant to homography. Moreover, it can be easily plugged into other popular CNN-based methods. Extensive experiments on the POT benchmark, S-COCO-Proj, and MNIST-Proj dataset show that our proposed method is effective for planar object tracking, homography estimation, and classification.
Estimating homography from an image pair is a fundamental problem in image alignment. Unsupervised learning methods have received increasing attention in this field due to their promising performance and label-free training. However, existing methods do not explicitly consider the problem of plane-induced parallax, which will make the predicted homography compromised on multiple planes. In this work, we propose a novel method HomoGAN to guide unsupervised homography estimation to focus on the dominant plane. First, a multi-scale transformer network is designed to predict homography from the feature pyramids of input images in a coarse-to-fine fashion. Moreover, we propose an unsupervised GAN to impose coplanarity constraint on the predicted homography, which is realized by using a generator to predict a mask of aligned regions, and then a discriminator to check if two masked feature maps are induced by a single homography. To validate the effectiveness of HomoGAN and its components, we conduct extensive experiments on a large-scale dataset, and the results show that our matching error is 22% lower than the previous SOTA method. Code is available at https://github.com/megvii-research/HomoGAN.
We propose Iterative Homography Network, namely IHN, a new deep homography estimation architecture. Different from previous works that achieve iterative refinement by network cascading or untrainable IC-LK iterator, the iterator of IHN has tied weights and is completely trainable. IHN achieves state-of-the-art accuracy on several datasets including challenging scenes. We propose 2 versions of IHN: (1) IHN for static scenes, (2) IHN-mov for dynamic scenes with moving objects. Both versions can be arranged in 1-scale for efficiency or 2-scale for accuracy. We show that the basic 1-scale IHN already outperforms most of the existing methods. On a variety of datasets, the 2-scale IHN outperforms all competitors by a large gap. We introduce IHN-mov by producing an inlier mask to further improve the estimation accuracy of moving-objects scenes. We experimentally show that the iterative framework of IHN can achieve 95% error reduction while considerably saving network parameters. When processing sequential image pairs, IHN can achieve 32.7 fps, which is about 8x the speed of IC-LK iterator. Source code is available at https://github.com/imdumpl78/IHN.
Local feature matching is a computationally intensive task at the subpixel level. While detector-based methods coupled with feature descriptors struggle in low-texture scenes, CNN-based methods with a sequential extract-to-match pipeline, fail to make use of the matching capacity of the encoder and tend to overburden the decoder for matching. In contrast, we propose a novel hierarchical extract-and-match transformer, termed as MatchFormer. Inside each stage of the hierarchical encoder, we interleave self-attention for feature extraction and cross-attention for feature matching, yielding a human-intuitive extract-and-match scheme. Such a match-aware encoder releases the overloaded decoder and makes the model highly efficient. Further, combining self- and cross-attention on multi-scale features in a hierarchical architecture improves matching robustness, particularly in low-texture indoor scenes or with less outdoor training data. Thanks to such a strategy, MatchFormer is a multi-win solution in efficiency, robustness, and precision. Compared to the previous best method in indoor pose estimation, our lite MatchFormer has only 45% GFLOPs, yet achieves a +1.3% precision gain and a 41% running speed boost. The large MatchFormer reaches state-of-the-art on four different benchmarks, including indoor pose estimation (ScanNet), outdoor pose estimation (MegaDepth), homography estimation and image matching (HPatch), and visual localization (InLoc).
Efficient detection and description of geometric regions in images is a prerequisite in visual systems for localization and mapping. Such systems still rely on traditional hand-crafted methods for efficient generation of lightweight descriptors, a common limitation of the more powerful neural network models that come with high compute and specific hardware requirements. In this paper, we focus on the adaptations required by detection and description neural networks to enable their use in computationally limited platforms such as robots, mobile, and augmented reality devices. To that end, we investigate and adapt network quantization techniques to accelerate inference and enable its use on compute limited platforms. In addition, we revisit common practices in descriptor quantization and propose the use of a binary descriptor normalization layer, enabling the generation of distinctive binary descriptors with a constant number of ones. ZippyPoint, our efficient quantized network with binary descriptors, improves the network runtime speed, the descriptor matching speed, and the 3D model size, by at least an order of magnitude when compared to full-precision counterparts. These improvements come at a minor performance degradation as evaluated on the tasks of homography estimation, visual localization, and map-free visual relocalization. Code and models are available at https://github.com/menelaoskanakis/ZippyPoint.
The homography matrix is a key component in various vision-based robotic tasks. Traditionally, homography estimation algorithms are classified into feature- or intensity-based. The main advantages of the latter are their versatility, accuracy, and robustness to arbitrary illumination changes. On the other hand, they have a smaller domain of convergence than the feature-based solutions. Their combination is hence promising, but existing techniques only apply them sequentially. This paper proposes a new hybrid method that unifies both classes into a single nonlinear optimization procedure, applies the same minimization method, and uses the same homography parametrization and warping function. Experimental validation using a classical testing framework shows that the proposed unified approach has improved convergence properties compared to each individual class. These are also demonstrated in a visual tracking application. As a final contribution, our ready-to-use implementation of the algorithm is made publicly available to the research community.
Structure from motion (SFM) and ground plane homography estimation are critical to autonomous driving and other robotics applications. Recently, much progress has been made in using deep neural networks for SFM and homography estimation respectively. However, directly applying existing methods for ground plane homography estimation may fail because the road is often a small part of the scene. Besides, the performances of deep SFM approaches are still inferior to traditional methods. In this paper, we propose a method that learns to solve both problems in an end-to-end manner, improving performance on both. The proposed networks consist of a Depth-CNN, a Pose-CNN and a Ground-CNN. The Depth-CNN and Pose-CNN estimate dense depth map and ego-motion respectively, solving SFM, while the Pose-CNN and Ground-CNN followed by a homography layer solve the ground plane estimation problem. By enforcing coherency between SFM and homography estimation results, the whole network can be trained end to end using photometric loss and homography loss without any groundtruth except the road segmentation provided by an off-the-shelf segmenter. Comprehensive experiments are conducted on KITTI benchmark to demonstrate promising results compared with various state-of-the-art approaches.