Search

The task of 3D single object tracking (SOT) with LiDAR point clouds is crucial for various applications, such as autonomous driving and robotics. However, existing approaches have primarily relied on appearance matching or motion modeling within only two successive frames, thereby overlooking the long-range continuous motion property of objects in 3D space. To address this issue, this paper presents a novel approach that views each tracklet as a continuous stream: at each timestamp, only the current frame is fed into the network to interact with multi-frame historical features stored in a memory bank, enabling efficient exploitation of sequential information. To achieve effective cross-frame message passing, a hybrid attention mechanism is designed to account for both long-range relation modeling and local geometric feature extraction. Furthermore, to enhance the utilization of multi-frame features for robust tracking, a contrastive sequence enhancement strategy is designed, which uses ground truth tracklets to augment training sequences and promote discrimination against false positives in a contrastive manner. Extensive experiments demonstrate that the proposed method outperforms the state-of-the-art method by significant margins (approximately 8%, 6%, and 12% improvements in the success performance on KITTI, nuScenes, and Waymo, respectively).

With the prevalence of LiDAR sensors in autonomous driving, 3D object tracking has received increasing attention. In a point cloud sequence, 3D object tracking aims to predict the location and orientation of an object in consecutive frames given an object template. Motivated by the success of transformers, we propose Point Tracking TRansformer (PTTR), which efficiently predicts high-quality 3D tracking results in a coarse-to-fine manner with the help of transformer operations. PTTR consists of three novel designs. 1) Instead of random sampling, we design Relation-Aware Sampling to preserve relevant points to the given template during subsampling. 2) We propose a Point Relation Transformer for effective feature aggregation and feature matching between the template and search region. 3) Based on the coarse tracking results, we employ a novel Prediction Refinement Module to obtain the final refined prediction through local feature pooling. In addition, motivated by the favorable properties of the Bird's-Eye View (BEV) of point clouds in capturing object motion, we further design a more advanced framework named PTTR++, which incorporates both the point-wise view and BEV representation to exploit their complementary effect in generating high-quality tracking results. PTTR++ substantially boosts the tracking performance on top of PTTR with low computational overhead. Extensive experiments over multiple datasets show that our proposed approaches achieve superior 3D tracking accuracy and efficiency., Comment: arXiv admin note: substantial text overlap with arXiv:2112.02857

3D object detection using point clouds has attracted increasing attention due to its wide applications in autonomous driving and robotics. However, most existing studies focus on single point cloud frames without harnessing the temporal information in point cloud sequences. In this paper, we design TransPillars, a novel transformer-based feature aggregation technique that exploits temporal features of consecutive point cloud frames for multi-frame 3D object detection. TransPillars aggregates spatial-temporal point cloud features from two perspectives. First, it fuses voxel-level features directly from multi-frame feature maps instead of pooled instance features to preserve instance details with contextual information that are essential to accurate object localization. Second, it introduces a hierarchical coarse-to-fine strategy to fuse multi-scale features progressively to effectively capture the motion of moving objects and guide the aggregation of fine features. Besides, a variant of deformable transformer is introduced to improve the effectiveness of cross-frame feature matching. Extensive experiments show that our proposed TransPillars achieves state-of-art performance as compared to existing multi-frame detection approaches. Code will be released.

3D object detection with surround-view images is an essential task for autonomous driving. In this work, we propose DETR4D, a Transformer-based framework that explores sparse attention and direct feature query for 3D object detection in multi-view images. We design a novel projective cross-attention mechanism for query-image interaction to address the limitations of existing methods in terms of geometric cue exploitation and information loss for cross-view objects. In addition, we introduce a heatmap generation technique that bridges 3D and 2D spaces efficiently via query initialization. Furthermore, unlike the common practice of fusing intermediate spatial features for temporal aggregation, we provide a new perspective by introducing a novel hybrid approach that performs cross-frame fusion over past object queries and image features, enabling efficient and robust modeling of temporal information. Extensive experiments on the nuScenes dataset demonstrate the effectiveness and efficiency of the proposed DETR4D.

Adversarial training (AT) is one of the most effective ways for improving the robustness of deep convolution neural networks (CNNs). Just like common network training, the effectiveness of AT relies on the design of basic network components. In this paper, we conduct an in-depth study on the role of the basic ReLU activation component in AT for robust CNNs. We find that the spatially-shared and input-independent properties of ReLU activation make CNNs less robust to white-box adversarial attacks with either standard or adversarial training. To address this problem, we extend ReLU to a novel Sparta activation function (Spatially attentive and Adversarially Robust Activation), which enables CNNs to achieve both higher robustness, i.e., lower error rate on adversarial examples, and higher accuracy, i.e., lower error rate on clean examples, than the existing state-of-the-art (SOTA) activation functions. We further study the relationship between Sparta and the SOTA activation functions, providing more insights about the advantages of our method. With comprehensive experiments, we also find that the proposed method exhibits superior cross-CNN and cross-dataset transferability. For the former, the adversarially trained Sparta function for one CNN (e.g., ResNet-18) can be fixed and directly used to train another adversarially robust CNN (e.g., ResNet-34). For the latter, the Sparta function trained on one dataset (e.g., CIFAR-10) can be employed to train adversarially robust CNNs on another dataset (e.g., SVHN). In both cases, Sparta leads to CNNs with higher robustness than the vanilla ReLU, verifying the flexibility and versatility of the proposed method., 25 pages, 5 figures

Deep learning-based 3D object detection has achieved unprecedented success with the advent of large-scale autonomous driving datasets. However, drastic performance degradation remains a critical challenge for cross-domain deployment. In addition, existing 3D domain adaptive detection methods often assume prior access to the target domain annotations, which is rarely feasible in the real world. To address this challenge, we study a more realistic setting, unsupervised 3D domain adaptive detection, which only utilizes source domain annotations. 1) We first comprehensively investigate the major underlying factors of the domain gap in 3D detection. Our key insight is that geometric mismatch is the key factor of domain shift. 2) Then, we propose a novel and unified framework, Multi-Level Consistency Network (MLC-Net), which employs a teacher-student paradigm to generate adaptive and reliable pseudo-targets. MLC-Net exploits point-, instance- and neural statistics-level consistency to facilitate cross-domain transfer. Extensive experiments demonstrate that MLC-Net outperforms existing state-of-the-art methods (including those using additional target domain information) on standard benchmarks. Notably, our approach is detector-agnostic, which achieves consistent gains on both single- and two-stage 3D detectors.

Standards play an important and unreplaceable role in everywhere of human life and social activities. Most of countries pay much attention to standards development and application. However, all of standards are still paper media and mainly for human reading. It greatly restricts the efficiency of usage, broadcast, and data interchange especially in such digital era with computer being widely used. In this pater, we discuss a new form of standard, which is machine readable, understandable, executable, testable, and applicable as well as human applying.

Since many modern multi-functional radars emitting dense complex signals are applied in the complex electromagnetic environment, it is challenge to classify and identify the radar emitters in the electronic reconnaissance field. In order to classify and identify radar emitters accurately and rapidly, a kind of standard adaptive extended radar emitter modelling structure and the complete process of obtaining radar emitter model parameters are proposed in this study. The simulation analysis results show that the proposed standard modelling structure and complete process can rapidly realise the accurate classification of many multi-functional radar emitters; hence can further benefit to accurate identification of emitters.

Two well-preserved fossil dragonflies from the Late Mesozoic Yixian Formation, Liaoning Province, China are described and assigned to a new genus, Sopholibellula gen. nov. in Araripelibellulidae Bechly, 1996, closely related to the type genus Araripelibellula. This new genus differs from Araripelibellula in the following characters: origins of RP and MA distinctly separated at arculus in both pairs of wings; anal loop wider and shorter, with Y-shaped veins inside; MA and IR2 not zigzag; several small intercalary veins present in the postdiscoidal area of hindwing; cells smaller and much more dense, especially in the apex and hind margin; bigger in size. Structures, including head, abdomen and parts of legs, were first described in details of this family.

Effects of temperature on population parameters and the intrinsic rate of natural increase of the leafminer, Liriomyza sativae Blanchard, were studied at constant temperatures, 15, 20, 25, 30 and 3593 80% RH and a photoperiod of 12:12 (L:D) in the laboratory with Phaseolus vulgaris as the host plant. Developmental time of the immature stage decreased from 38 d at 15C to lld at 3593 Regression equations relating temperature (t) to development rates (y) for egg, larval and pupal stages, were y = 1.7862t - 13.841, y = 1.162t - 4.946 and y= 0.634t - 5.146, respectively. Longevity of female adult decreased from 20 d at 15°C to 9 d with temperature up to 35C The most favorable temperature range for reproduction was 20°C - 30oC in which the fecundity ranged from 158 to 282 eggs per female. The lowest total mortality was 9% at 25oC and the highest was 49% at 35 93 High intrinsic rate of natural increase (rm) was 0.27 and high net reproductive rate (R0) was 116.8 at temperature range between 25 t and 30 t indicating that this range was optimal for population growth and that population density might increase 117 times per generation under this temperature condition. Mean generation time (T) and time for population to double (t) decreased as temperature increased, showing a negative linear trend with temperature. The relationship between finite rate of increase (A) and temperature, however, was a positive linear regression.

The developmental thresholds and number of degree-days needed to complete development were determined for the immature stages and the preovipositional period of the pumpkin fly, Bactrocera tau at 6 constant temperatures (19, 22, 25, 28, 31 and 34 °C). They are (threshold/degree-days) 9.32 °C/15.75 for the eggs, 6.67°C/113.82 for the larvae, 9.30°C/144.35 for the pupae and 8.56°C/180.8 for the preovipositional period. Developmental rate was related to temperature by a sigmoid curve. Gross and net fecundity, and fertility were greatest at 25 °C, although daily fecundity and fertility were greatest at 28 °C. Intrinsic and finite rates of population increase were greatest at 28 °C. Zusammenfassung Uber die Wirkung der Temperatur auf das Populationswachstum von Bactrocera tau (Walk.) (Dipt., Tephritidae) Es wurden der Entwicklungsnullpunkt und die Zahl der Tagesgrade, die zur Vollendung der Entwicklung notig sind, fur die Entwicklungsstadien und die Praovisitions-Periode der Kurbisfruchtfliege, Bactrocera tau bei 6 konstanten Temperaturen (19, 22, 25, 28, 31, 34 °C) untersucht. Der Entwicklungsnullpunkt und die Tagesgrade betrugen 9,3°C/15.75 fur die Eier, 6,7°C/113,8 fur die Larven, 9,3°C/144,4 fur die Puppen und 8,6 °C/180,8 fur die Praovipositionsperiode. Das Verhaltnis von Entwicklungsrate zur Temperatur wurde durch eine Simoid-Kurve gekennzeichnet. Die Brutto-und Netto-Fekunditat und Fertilitat waren bei 25 °C am hochsten, obgleich die tagliche Fekunditat und Fertilitat bei 28 °C ihren hochsten Wert erreichten. Die Intrinsic-Endrate des Populationswachstums war bei 28 °C am grosten.