Siamese network based trackers formulate tracking as convolutional feature cross-correlation between target template and searching region. However, Siamese trackers still have accuracy gap compared with state-of-the-art algorithms and they cannot take advantage of feature from deep networks, such as ResNet-50 or deeper. In this work we prove the core reason comes from the lack of strict translation invariance. By comprehensive theoretical analysis and experimental validations, we break this restriction through a simple yet effective spatial aware sampling strategy and successfully train a ResNet-driven Siamese tracker with significant performance gain. Moreover, we propose a new model architecture to perform depth-wise and layer-wise aggregations, which not only further improves the accuracy but also reduces the model size. We conduct extensive ablation studies to demonstrate the effectiveness of the proposed tracker, which obtains currently the best results on four large tracking benchmarks, including OTB2015, VOT2018, UAV123, and LaSOT. Our model will be released to facilitate further studies based on this problem.

/pdf/siamrpn-evolution-of-siamese-visual-tracking-with-very-deep-39kz6y4yw2.pdf

SiamRPN++: Evolution of Siamese Visual Tracking With Very Deep Networks

We introduce here a large tracking database that offers an unprecedentedly wide coverage of common moving objects in the wild, called GOT-10k. Specifically, GOT-10k is built upon the backbone of WordNet structure [1] and it populates the majority of over 560 classes of moving objects and 87 motion patterns, magnitudes wider than the most recent similar-scale counterparts [19] , [20] , [23] , [26] . By releasing the large high-diversity database, we aim to provide a unified training and evaluation platform for the development of class-agnostic, generic purposed short-term trackers. The features of GOT-10k and the contributions of this article are summarized in the following. (1) GOT-10k offers over 10,000 video segments with more than 1.5 million manually labeled bounding boxes, enabling unified training and stable evaluation of deep trackers. (2) GOT-10k is by far the first video trajectory dataset that uses the semantic hierarchy of WordNet to guide class population, which ensures a comprehensive and relatively unbiased coverage of diverse moving objects. (3) For the first time, GOT-10k introduces the one-shot protocol for tracker evaluation, where the training and test classes are zero-overlapped . The protocol avoids biased evaluation results towards familiar objects and it promotes generalization in tracker development. (4) GOT-10k offers additional labels such as motion classes and object visible ratios, facilitating the development of motion-aware and occlusion-aware trackers. (5) We conduct extensive tracking experiments with 39 typical tracking algorithms and their variants on GOT-10k and analyze their results in this paper. (6) Finally, we develop a comprehensive platform for the tracking community that offers full-featured evaluation toolkits, an online evaluation server, and a responsive leaderboard. The annotations of GOT-10k’s test data are kept private to avoid tuning parameters on it.

GOT-10k: A Large High-Diversity Benchmark for Generic Object Tracking in the Wild

Mobile 3D mapping for surveying earthwork projects using an Unmanned Aerial Vehicle (UAV) system

State of research in automatic as-built modelling

Research on unmanned aerial vehicles (UAV) has been increasingly popular in the past decades, and UAVs have been widely used in industrial inspection, remote sensing for mapping & surveying, rescui...

https://www.tandfonline.com/doi/pdf/10.1080/10095020.2017.1420509

A survey on vision-based UAV navigation

Not all measured features in SLAM/SfM contribute to accurate localization during the estimation process, thus it is sensible to utilize only those that do. This paper describes a method for selecting a subset of features that are of high utility for localization in the SLAM/SfM estimation process. It is derived by examining the observability of SLAM and, being complimentary to the estimation process, it easily integrates into existing SLAM systems. The measure of estimation utility is formulated with temporal and instantaneous observability indices. Efficient computation strategies for the observability indices are described based on incremental singular value decomposition (SVD) and greedy selection for the temporal and instantaneous observability indices, respectively. The greedy selection is near-optimal since the observability index is (approximately) submodular. The proposed method improves localization and data association. Controlled synthetic experiments with ground truth demonstrate the improved localization accuracy, and real-time SLAM experiments demonstrate the improved data association.

/pdf/good-features-to-track-for-visual-slam-2s994v6flm.pdf

Good features to track for visual SLAM

Currently, much of the manual labor needed to generate as-built building information models (BIMs) of existing facilities is spent converting raw point cloud data sets (PCDs) to BIM descriptions. Automating the PCD conversion process can drastically reduce the cost of generating as-built BIMs. Due to the widespread existence of planar structures in civil infrastructures, detecting and extracting planar patches from raw PCDs is a fundamental step in the conversion pipeline from PCDs to BIMs. However, existing methods cannot effectively address both automatically detecting and extracting planar patches from infrastructure PCDs. The existing methods cannot resolve the problem due to the large scale and model complexity of civil infrastructure, or due to the requirements of extra constraints or known information. To address the problem, this article presents a novel framework for automatically detecting and extracting planar patches from large-scale and noisy raw PCDs. The proposed method automatically detects planar structures, estimates the parametric plane models, and determines the boundaries of the planar patches. The first step recovers existing linear dependence relationships amongst points in the PCD by solving a group-sparsity inducing optimization problem. Next, a spectral clustering procedure based on the recovered linear dependence relationships segments the PCD. Then, for each segmented group, model parameters of the extracted planes are estimated via singular value decomposition (SVD) and maximum likelihood estimation sample consensus (MLESAC). Finally, the α-shape algorithm detects the boundaries of planar structures based on a projection of the data to the planar model. The proposed approach is evaluated comprehensively by experiments on two types of PCDs from real-world infrastructures, one captured directly by laser scanners and the other reconstructed from video using structure-from-motion techniques. To evaluate the performance comprehensively, five evaluation metrics are proposed which measure different aspects of performance. Experimental results reveal that the proposed method outperforms the existing methods, in the sense that the method automatically and accurately extracts planar patches from large-scaled raw PCDs without any extra constraints nor user assistance.

/pdf/a-sparsity-inducing-optimization-based-algorithm-for-planar-44e9kbrorl.pdf

A Sparsity-Inducing Optimization-Based Algorithm for Planar Patches Extraction from Noisy Point-Cloud Data

This paper proposes a simple yet effective approach to learn visual features online for improving loop-closure detection and place recognition, based on bag-of-words frameworks. The approach learns a codeword in the bag-of-words model from a pair of matched features from two consecutive frames, such that the codeword has temporally-derived perspective invariance to camera motion. The learning algorithm is efficient: the binary descriptor is generated from the mean image patch, and the mask is learned based on discriminative projection by minimizing the intra-class distances among the learned feature and the two original features. A codeword is generated by packaging the learned descriptor and mask, with a masked Hamming distance defined to measure the distance between two codewords. The geometric properties of the learned codewords are then mathematically justified. In addition, hypothesis constraints are imposed through temporal consistency in matched codewords, which improves precision. The approach, integrated in an incremental bag-of-words system, is validated on multiple benchmark data sets and compared to state-of-the-art methods. Experiments demonstrate improved precision/recall outperforming state of the art with little loss in runtime.

Learning binary features online from motion dynamics for incremental loop-closure detection and place recognition

/pdf/robust-feature-detection-acquisition-and-tracking-for-4f4znnrtl6.pdf

Robust Feature Detection, Acquisition and Tracking for Relative Navigation in Space with a Known Target

Converting remote sensing point cloud data (PCD) into solid CAD models consisting of civil infrastructure components is a crucial step in generating the asbuilt building information models. Previous research has enabled automatic generation of surface primitives from raw PCDs. However, the fully automatic conversion from surface primitives to infrastructure component models remains an unsolved problem. In this work, an automatic and linear-runtime approach is presented which generates the as-built infrastructure component models by recognizing the solid CAD entities and learning the infrastructure component labels from the fitted surface primitives. The algorithm utilizes a decision tree with the following decision variables: the type, parametric model, orientation, and mutual geometric relations of the fitted surface primitives. The decision tree is trained with easily generated synthetic data and is applied to query real-world data with complexity O(1). The output of the solid entities includes cuboid, cylinder, ball, etc. and the infrastructure component labels such as columns, caps, deck, beams, etc. The algorithm is tested with various PCDs modeling real bridges.

Guangcong Zhang

Papers

Good features to track for visual SLAM

A Sparsity-Inducing Optimization-Based Algorithm for Planar Patches Extraction from Noisy Point-Cloud Data

Learning binary features online from motion dynamics for incremental loop-closure detection and place recognition

Robust Feature Detection, Acquisition and Tracking for Relative Navigation in Space with a Known Target

Automatic Generation of As-Built Geometric Civil Infrastructure Models from Point Cloud Data