In order to get raw images of high quality for downstream Image Signal Process (ISP), in this paper we present an Efficient Locally Multiplicative Transformer called ELMformer for raw image restoration. ELMformer contains two core designs especially for raw images whose primitive attribute is single-channel. The first design is a Bi-directional Fusion Projection (BFP) module, where we consider both the color characteristics of raw images and spatial structure of single-channel. The second one is that we propose a Locally Multiplicative Self-Attention (L-MSA) scheme to effectively deliver information from the local space to relevant parts. ELMformer can efficiently reduce the computational consumption and perform well on raw image restoration tasks. Enhanced by these two core designs, ELMformer achieves the highest performance and keeps the lowest FLOPs on raw denoising and raw deblurring benchmarks compared with state-of-the-arts. Extensive experiments demonstrate the superiority and generalization ability of ELMformer. On SIDD benchmark, our method has even better denoising performance than ISP-based methods which need huge amount of additional sRGB training images. The codes are release at https://github.com/leonmakise/ELMformer.
We present MapTR, a structured end-to-end framework for efficient online vectorized HD map construction. We propose a unified permutation-based modeling approach, i.e., modeling map element as a point set with a group of equivalent permutations, which avoids the definition ambiguity of map element and eases learning. We adopt a hierarchical query embedding scheme to flexibly encode structured map information and perform hierarchical bipartite matching for map element learning. MapTR achieves the best performance and efficiency among existing vectorized map construction approaches on nuScenes dataset. In particular, MapTR-nano runs at real-time inference speed ($25.1$ FPS) on RTX 3090, $8\times$ faster than the existing state-of-the-art camera-based method while achieving $3.3$ higher mAP. MapTR-tiny significantly outperforms the existing state-of-the-art multi-modality method by $13.5$ mAP while being faster. Qualitative results show that MapTR maintains stable and robust map construction quality in complex and various driving scenes. Abundant demos are available at \url{https://github.com/hustvl/MapTR} to prove the effectiveness in real-world scenarios. MapTR is of great application value in autonomous driving. Code will be released for facilitating further research and application.
Unlike the conventional Knowledge Distillation (KD), Self-KD allows a network to learn knowledge from itself without any guidance from extra networks. This paper proposes to perform Self-KD from image Mixture (MixSKD), which integrates these two techniques into a unified framework. MixSKD mutually distills feature maps and probability distributions between the random pair of original images and their mixup images in a meaningful way. Therefore, it guides the network to learn cross-image knowledge by modelling supervisory signals from mixup images. Moreover, we construct a self-teacher network by aggregating multi-stage feature maps for providing soft labels to supervise the backbone classifier, further improving the efficacy of self-boosting. Experiments on image classification and transfer learning to object detection and semantic segmentation demonstrate that MixSKD outperforms other state-of-the-art Self-KD and data augmentation methods. The code is available at https://github.com/winycg/Self-KD-Lib.
Spiking neural networks (SNNs) have attracted extensive attentions in Brain-inspired Artificial Intelligence and computational neuroscience. They can be used to simulate biological information processing in the brain at multiple scales. More importantly, SNNs serve as an appropriate level of abstraction to bring inspirations from brain and cognition to Artificial Intelligence. In this paper, we present the Brain-inspired Cognitive Intelligence Engine (BrainCog) for creating brain-inspired AI and brain simulation models. BrainCog incorporates different types of spiking neuron models, learning rules, brain areas, etc., as essential modules provided by the platform. Based on these easy-to-use modules, BrainCog supports various brain-inspired cognitive functions, including Perception and Learning, Decision Making, Knowledge Representation and Reasoning, Motor Control, and Social Cognition. These brain-inspired AI models have been effectively validated on various supervised, unsupervised, and reinforcement learning tasks, and they can be used to enable AI models to be with multiple brain-inspired cognitive functions. For brain simulation, BrainCog realizes the function simulation of decision-making, working memory, the structure simulation of the Neural Circuit, and whole brain structure simulation of Mouse brain, Macaque brain, and Human brain. An AI engine named BORN is developed based on BrainCog, and it demonstrates how the components of BrainCog can be integrated and used to build AI models and applications. To enable the scientific quest to decode the nature of biological intelligence and create AI, BrainCog aims to provide essential and easy-to-use building blocks, and infrastructural support to develop brain-inspired spiking neural network based AI, and to simulate the cognitive brains at multiple scales. The online repository of BrainCog can be found at https://github.com/braincog-x.
In this work, we propose PolarBEV for vision-based uneven BEV representation learning. To adapt to the foreshortening effect of camera imaging, we rasterize the BEV space both angularly and radially, and introduce polar embedding decomposition to model the associations among polar grids. Polar grids are rearranged to an array-like regular representation for efficient processing. Besides, to determine the 2D-to-3D correspondence, we iteratively update the BEV surface based on a hypothetical plane, and adopt height-based feature transformation. PolarBEV keeps real-time inference speed on a single 2080Ti GPU, and outperforms other methods for both BEV semantic segmentation and BEV instance segmentation. Thorough ablations are presented to validate the design. The code will be released at \url{https://github.com/SuperZ-Liu/PolarBEV}.
3D detection based on surround-view camera system is a critical technique in autopilot. In this work, we present Polar Parametrization for 3D detection, which reformulates position parametrization, velocity decomposition, perception range, label assignment and loss function in polar coordinate system. Polar Parametrization establishes explicit associations between image patterns and prediction targets, exploiting the view symmetry of surround-view cameras as inductive bias to ease optimization and boost performance. Based on Polar Parametrization, we propose a surround-view 3D DEtection TRansformer, named PolarDETR. PolarDETR achieves promising performance-speed trade-off on different backbone configurations. Besides, PolarDETR ranks 1st on the leaderboard of nuScenes benchmark in terms of both 3D detection and 3D tracking at the submission time (Mar. 4th, 2022). Code will be released at \url{https://github.com/hustvl/PolarDETR}.
The query mechanism introduced in the DETR method is changing the paradigm of object detection and recently there are many query-based methods have obtained strong object detection performance. However, the current query-based detection pipelines suffer from the following two issues. Firstly, multi-stage decoders are required to optimize the randomly initialized object queries, incurring a large computation burden. Secondly, the queries are fixed after training, leading to unsatisfying generalization capability. To remedy the above issues, we present featurized object queries predicted by a query generation network in the well-established Faster R-CNN framework and develop a Featurized Query R-CNN. Extensive experiments on the COCO dataset show that our Featurized Query R-CNN obtains the best speed-accuracy trade-off among all R-CNN detectors, including the recent state-of-the-art Sparse R-CNN detector. The code is available at {https://github.com/hustvl/Featurized-QueryRCNN.
Learning Bird's Eye View (BEV) representation from surrounding-view cameras is of great importance for autonomous driving. In this work, we propose a Geometry-guided Kernel Transformer (GKT), a novel 2D-to-BEV representation learning mechanism. GKT leverages the geometric priors to guide the transformer to focus on discriminative regions and unfolds kernel features to generate BEV representation. For fast inference, we further introduce a look-up table (LUT) indexing method to get rid of the camera's calibrated parameters at runtime. GKT can run at $72.3$ FPS on 3090 GPU / $45.6$ FPS on 2080ti GPU and is robust to the camera deviation and the predefined BEV height. And GKT achieves the state-of-the-art real-time segmentation results, i.e., 38.0 mIoU (100m$\times$100m perception range at a 0.5m resolution) on the nuScenes val set. Given the efficiency, effectiveness, and robustness, GKT has great practical values in autopilot scenarios, especially for real-time running systems. Code and models will be available at \url{https://github.com/hustvl/GKT}.
The estimation of cumulative distribution functions (CDF) is an important learning task with a great variety of downstream applications, e.g., risk assessments in predictions and decision making. We study functional regression of contextual CDFs where each data point is sampled from a linear combination of context dependent CDF bases. We propose estimation methods that estimate CDFs accurately everywhere. In particular, given $n$ samples with $d$ bases, we show estimation error upper bounds of $\widetilde O(\sqrt{d/n})$ for fixed design, random design, and adversarial context cases. We also derive matching information theoretic lower bounds, establishing minimax optimality for CDF functional regression. To complete our study, we consider agnostic settings where there is a mismatch in the data generation process. We characterize the error of the proposed estimator in terms of the mismatched error, and show that the estimator is well-behaved under model mismatch.
This paper introduces contrastive siamese (c-siam) network, an architecture for leveraging unlabeled acoustic data in speech recognition. c-siam is the first network that extracts high-level linguistic information from speech by matching outputs of two identical transformer encoders. It contains augmented and target branches which are trained by: (1) masking inputs and matching outputs with a contrastive loss, (2) incorporating a stop gradient operation on the target branch, (3) using an extra learnable transformation on the augmented branch, (4) introducing new temporal augment functions to prevent the shortcut learning problem. We use the Libri-light 60k unsupervised data and the LibriSpeech 100hrs/960hrs supervised data to compare c-siam and other best-performing systems. Our experiments show that c-siam provides 20% relative word error rate improvement over wav2vec baselines. A c-siam network with 450M parameters achieves competitive results compared to the state-of-the-art networks with 600M parameters.