Search

We explore Bird's-Eye View (BEV) generation, converting a BEV map into its corresponding multi-view street images. Valued for its unified spatial representation aiding multi-sensor fusion, BEV is pivotal for various autonomous driving applications. Creating accurate street-view images from BEV maps is essential for portraying complex traffic scenarios and enhancing driving algorithms. Concurrently, diffusion-based conditional image generation models have demonstrated remarkable outcomes, adept at producing diverse, high-quality, and condition-aligned results. Nonetheless, the training of these models demands substantial data and computational resources. Hence, exploring methods to fine-tune these advanced models, like Stable Diffusion, for specific conditional generation tasks emerges as a promising avenue. In this paper, we introduce a practical framework for generating images from a BEV layout. Our approach comprises two main components: the Neural View Transformation and the Street Image Generation. The Neural View Transformation phase converts the BEV map into aligned multi-view semantic segmentation maps by learning the shape correspondence between the BEV and perspective views. Subsequently, the Street Image Generation phase utilizes these segmentations as a condition to guide a fine-tuned latent diffusion model. This finetuning process ensures both view and style consistency. Our model leverages the generative capacity of large pretrained diffusion models within traffic contexts, effectively yielding diverse and condition-coherent street view images., Comment: Accepted at International Conference on Robotics and Automation(ICRA)

Traditional deep learning relies on end-to-end backpropagation for training, but it suffers from drawbacks such as high memory consumption and not aligning with biological neural networks. Recent advancements have introduced locally supervised learning, which divides networks into modules with isolated gradients and trains them locally. However, this approach can lead to performance lag due to limited interaction between these modules, and the design of auxiliary networks occupies a certain amount of GPU memory. To overcome these limitations, we propose a novel model called HPFF that performs hierarchical locally supervised learning and patch-level feature computation on the auxiliary networks. Hierarchical Locally Supervised Learning (HiLo) enables the network to learn features at different granularity levels along their respective local paths. Specifically, the network is divided into two-level local modules: independent local modules and cascade local modules. The cascade local modules combine two adjacent independent local modules, incorporating both updates within the modules themselves and information exchange between adjacent modules. Patch Feature Fusion (PFF) reduces GPU memory usage by splitting the input features of the auxiliary networks into patches for computation. By averaging these patch-level features, it enhances the network's ability to focus more on those patterns that are prevalent across multiple patches. Furthermore, our method exhibits strong generalization capabilities and can be seamlessly integrated with existing techniques. We conduct experiments on CIFAR-10, STL-10, SVHN, and ImageNet datasets, and the results demonstrate that our proposed HPFF significantly outperforms previous approaches, consistently achieving state-of-the-art performance across different datasets. Our code is available at: https://github.com/Zeudfish/HPFF., Comment: Accepted by ECCV2024

Deep neural networks conventionally employ end-to-end backpropagation for their training process, which lacks biological credibility and triggers a locking dilemma during network parameter updates, leading to significant GPU memory use. Supervised local learning, which segments the network into multiple local blocks updated by independent auxiliary networks. However, these methods cannot replace end-to-end training due to lower accuracy, as gradients only propagate within their local block, creating a lack of information exchange between blocks. To address this issue and establish information transfer across blocks, we propose a Momentum Auxiliary Network (MAN) that establishes a dynamic interaction mechanism. The MAN leverages an exponential moving average (EMA) of the parameters from adjacent local blocks to enhance information flow. This auxiliary network, updated through EMA, helps bridge the informational gap between blocks. Nevertheless, we observe that directly applying EMA parameters has certain limitations due to feature discrepancies among local blocks. To overcome this, we introduce learnable biases, further boosting performance. We have validated our method on four image classification datasets (CIFAR-10, STL-10, SVHN, ImageNet), attaining superior performance and substantial memory savings. Notably, our method can reduce GPU memory usage by more than 45\% on the ImageNet dataset compared to end-to-end training, while achieving higher performance. The Momentum Auxiliary Network thus offers a new perspective for supervised local learning. Our code is available at: https://github.com/JunhaoSu0/MAN., Comment: Accepted by ECCV2024(Oral)

Monocular 3D object detection plays a crucial role in autonomous driving. However, existing monocular 3D detection algorithms depend on 3D labels derived from LiDAR measurements, which are costly to acquire for new datasets and challenging to deploy in novel environments. Specifically, this study investigates the pipeline for training a monocular 3D object detection model on a diverse collection of 3D and 2D datasets. The proposed framework comprises three components: (1) a robust monocular 3D model capable of functioning across various camera settings, (2) a selective-training strategy to accommodate datasets with differing class annotations, and (3) a pseudo 3D training approach using 2D labels to enhance detection performance in scenes containing only 2D labels. With this framework, we could train models on a joint set of various open 3D/2D datasets to obtain models with significantly stronger generalization capability and enhanced performance on new dataset with only 2D labels. We conduct extensive experiments on KITTI/nuScenes/ONCE/Cityscapes/BDD100K datasets to demonstrate the scaling ability of the proposed method.

The Hot Universe Baryon Surveyor (HUBS) is a proposed space-based X-ray telescope for detecting X-ray emissions from the hot gas content in our universe. With its unprecedented spatially-resolved high-resolution spectroscopy and large field of view, the HUBS mission will be uniquely qualified to measure the physical and chemical properties of the hot gas in the interstellar medium, the circumgalactic medium, the intergalactic medium, and the intracluster medium. These measurements will be valuable for two key scientific goals of HUBS, namely to unravel the AGN and stellar feedback physics that governs the formation and evolution of galaxies, and to probe the baryon budget and multi-phase states from galactic to cosmological scales. In addition to these two goals, the HUBS mission will also help us solve some problems in the fields of galaxy clusters, AGNs, diffuse X-ray backgrounds, supernova remnants, and compact objects. This paper discusses the perspective of advancing these fields using the HUBS telescope., Comment: 52 pages, 22 figures. Accepted for publication in Science China: Physics, Mechanics and Astronomy

Just like weights, bias terms are the learnable parameters of many popular machine learning models, including neural networks. Biases are thought to enhance the representational power of neural networks, enabling them to solve a variety of tasks in computer vision. However, we argue that biases can be disregarded for some image-related tasks such as image classification, by considering the intrinsic distribution of images in the input space and desired model properties from first principles. Our findings suggest that zero-bias neural networks can perform comparably to biased networks for practical image classification tasks. We demonstrate that zero-bias neural networks possess a valuable property called scalar (multiplication) invariance. This means that the prediction of the network remains unchanged when the contrast of the input image is altered. We extend scalar invariance to more general cases, enabling formal verification of certain convex regions of the input space. Additionally, we prove that zero-bias neural networks are fair in predicting the zero image. Unlike state-of-the-art models that may exhibit bias toward certain labels, zero-bias networks have uniform belief in all labels. We believe dropping bias terms can be considered as a geometric prior in designing neural network architecture for image classification, which shares the spirit of adapting convolutions as the transnational invariance prior. The robustness and fairness advantages of zero-bias neural networks may also indicate a promising path towards trustworthy and ethical AI., Comment: 22 pages, 8 figures

Having reliable specifications is an unavoidable challenge in achieving verifiable correctness, robustness, and interpretability of AI systems. Existing specifications for neural networks are in the paradigm of data as specification. That is, the local neighborhood centering around a reference input is considered to be correct (or robust). While existing specifications contribute to verifying adversarial robustness, a significant problem in many research domains, our empirical study shows that those verified regions are somewhat tight, and thus fail to allow verification of test set inputs, making them impractical for some real-world applications. To this end, we propose a new family of specifications called neural representation as specification, which uses the intrinsic information of neural networks - neural activation patterns (NAPs), rather than input data to specify the correctness and/or robustness of neural network predictions. We present a simple statistical approach to mining neural activation patterns. To show the effectiveness of discovered NAPs, we formally verify several important properties, such as various types of misclassifications will never happen for a given NAP, and there is no ambiguity between different NAPs. We show that by using NAP, we can verify a significant region of the input space, while still recalling 84% of the data on MNIST. Moreover, we can push the verifiable bound to 10 times larger on the CIFAR10 benchmark. Thus, we argue that NAPs can potentially be used as a more reliable and extensible specification for neural network verification., Comment: 19 pages, 16 figures

Caricature is an artistic representation that deliberately exaggerates the distinctive features of a human face to convey humor or sarcasm. However, reconstructing a 3D caricature from a 2D caricature image remains a challenging task, mostly due to the lack of data. We propose to fill this gap by introducing 3DCaricShop, the first large-scale 3D caricature dataset that contains 2000 high-quality diversified 3D caricatures manually crafted by professional artists. 3DCaricShop also provides rich annotations including a paired 2D caricature image, camera parameters and 3D facial landmarks. To demonstrate the advantage of 3DCaricShop, we present a novel baseline approach for single-view 3D caricature reconstruction. To ensure a faithful reconstruction with plausible face deformations, we propose to connect the good ends of the detailrich implicit functions and the parametric mesh representations. In particular, we first register a template mesh to the output of the implicit generator and iteratively project the registration result onto a pre-trained PCA space to resolve artifacts and self-intersections. To deal with the large deformation during non-rigid registration, we propose a novel view-collaborative graph convolution network (VCGCN) to extract key points from the implicit mesh for accurate alignment. Our method is able to generate highfidelity 3D caricature in a pre-defined mesh topology that is animation-ready. Extensive experiments have been conducted on 3DCaricShop to verify the significance of the database and the effectiveness of the proposed method., Comment: CVPR 2021. Project page:https://qiuyuda.github.io/3DCaricShop/

Using X-ray sources as sensitive probes of stellar dynamical interactions in globular clusters (GCs), we study the mass segregation and binary burning processes in $\omega$ Cen. We show that the mass segregation of X-ray sources is quenched in $\omega$ Cen, while the X-ray source abundance of $\omega$ Cen is much smaller than other GCs, and the binary hardness ratio (defined as $L_{\rm X}/(L_{\rm K}f_{b})$, with $f_{b}$ the binary fraction, $L_{\rm X}$ and $L_{\rm K}$ the cumulative X-ray and K band luminosity of GCs, respectively) of $\omega$ Cen is located far below the $L_{\rm X}/(L_{\rm K}f_{b})-\sigma_{c}$ correlation line of the dynamically normal GCs. These evidences suggest that the binary burning processes are highly suppressed in $\omega$ Cen, and other "heating mechanisms", very likely a black hole subsystem (BHS), are essential in the dynamical evolution of $\omega$ Cen. Through the "black hole burning" processes (i.e., dynamical hardening of the BH binaries), the BHS is dominating the energy production of $\omega$ Cen, which also makes $\omega$ Cen a promising factory of gravitational-wave sources in the Galaxy., Comment: 13 pages, 3 figures, 1 table. Accepted for publication in ApJ

Using archival {\it Chandra} observations with a total effective exposure of 323 ks, we derive an updated catalog of point sources in the bulge globular cluster M28. The catalog contains 502 X-ray sources within an area of $\sim475\, \rm arcmin^{2}$, and more than $90\%$ of these sources are first detected in this cluster. We find significant dips in the radial distribution profiles of X-ray sources in M28, with the projected distance and width of the distribution dip for bright ($L_{X} \gtrsim 4.5\times 10^{30} {\rm\ erg\ \,s^{-1}}$) X-ray sources are larger than the faint ($L_{X} \lesssim 4.5\times 10^{30} {\rm\ erg\ \,s^{-1}}$) sources. The "generalized King model" fitting give a slightly larger average mass for the bright sources ($1.30\pm0.15\,M_{\odot}$) than the faint sources ($1.09\pm0.14\,M_{\odot}$), which support a universal mass segregation delay between heavy objects in GCs. Compared with 47 Tuc and Terzan 5, we show that the dynamical age of M28 is comparable to Terzan 5 and much smaller than 47 Tuc, but it is evolving more fast (i.e., with smaller two-body relaxation timescale) than 47 Tuc. These features may suggest an acceleration effect of cluster dynamical evolution by tidal shock in M28. Besides, we find an abnormal deficiency of X-ray sources in the central region ($R \lesssim 1.5 \rm~arcmin$) of M28 than its outskirts, which indicate that M28 may have suffered an early phase of primordial binary disruption within its central region, and mass segregation effect will erase such a phenomenon as cluster evolve to older dynamical age., Comment: 18 pages, 5 figures, accepted for publication in ApJ