YOLO has become a central real-time object detection system for robotics, driverless cars, and video monitoring applications. We present a comprehensive analysis of YOLO's evolution, examining the innovations and contributions in each iteration from the original YOLO up to YOLOv8, YOLO-NAS, and YOLO with Transformers. We start by describing the standard metrics and postprocessing; then, we discuss the major changes in network architecture and training tricks for each model. Finally, we summarize the essential lessons from YOLO's development and provide a perspective on its future, highlighting potential research directions to enhance real-time object detection systems.
The recent detection transformer (DETR) has advanced object detection, but its application on resource-constrained devices requires massive computation and memory resources. Quantization stands out as a solution by representing the network in low-bit parameters and operations. However, there is a significant performance drop when performing low-bit quantized DETR (Q-DETR) with existing quantization methods. We find that the bottlenecks of Q-DETR come from the query information distortion through our empirical analyses. This paper addresses this problem based on a distribution rectification distillation (DRD). We formulate our DRD as a bi-level optimization problem, which can be derived by generalizing the information bottleneck (IB) principle to the learning of Q-DETR. At the inner level, we conduct a distribution alignment for the queries to maximize the self-information entropy. At the upper level, we introduce a new foreground-aware query matching scheme to effectively transfer the teacher information to distillation-desired features to minimize the conditional information entropy. Extensive experimental results show that our method performs much better than prior arts. For example, the 4-bit Q-DETR can theoretically accelerate DETR with ResNet-50 backbone by 6.6x and achieve 39.4% AP, with only 2.6% performance gaps than its real-valued counterpart on the COCO dataset.
Safety is the primary priority of autonomous driving. Nevertheless, no published dataset currently supports the direct and explainable safety evaluation for autonomous driving. In this work, we propose DeepAccident, a large-scale dataset generated via a realistic simulator containing diverse accident scenarios that frequently occur in real-world driving. The proposed DeepAccident dataset includes 57K annotated frames and 285K annotated samples, approximately 7 times more than the large-scale nuScenes dataset with 40k annotated samples. In addition, we propose a new task, end-to-end motion and accident prediction, which can be used to directly evaluate the accident prediction ability for different autonomous driving algorithms. Furthermore, for each scenario, we set four vehicles along with one infrastructure to record data, thus providing diverse viewpoints for accident scenarios and enabling V2X (vehicle-to-everything) research on perception and prediction tasks. Finally, we present a baseline V2X model named V2XFormer that demonstrates superior performance for motion and accident prediction and 3D object detection compared to the single-vehicle model.
Cross-domain pulmonary nodule detection suffers from performance degradation due to a large shift of data distributions between the source and target domain. Besides, considering the high cost of medical data annotation, it is often assumed that the target images are unlabeled. Existing approaches have made much progress for this unsupervised domain adaptation setting. However, this setting is still rarely plausible in medical applications since the source medical data are often not accessible due to privacy concerns. This motivates us to propose a Source-free Unsupervised cross-domain method for Pulmonary nodule detection (SUP), named Instance-level Contrastive Instruction fine-tuning framework (ICI). It first adapts the source model to the target domain by utilizing instance-level contrastive learning. Then the adapted model is trained in a teacher-student interaction manner, and a weighted entropy loss is incorporated to further improve the accuracy. We establish a benchmark by adapting a pre-trained source model to three popular datasets for pulmonary nodule detection. To the best of our knowledge, this represents the first exploration of source-free unsupervised domain adaptation in medical image object detection. Our extensive evaluations reveal that SUP-ICI substantially surpasses existing state-of-the-art approaches, achieving FROC score improvements ranging from 8.98% to 16.05%. This breakthrough not only sets a new precedent for domain adaptation techniques in medical imaging but also significantly advances the field toward overcoming challenges posed by data privacy and availability. Code: https://github.com/Ruixxxx/SFUDA.
In recent years, transformer-based detectors have demonstrated remarkable performance in 2D visual perception tasks. However, their performance in multi-view 3D object detection remains inferior to the state-of-the-art (SOTA) of convolutional neural network based detectors. In this work, we investigate this issue from the perspective of bird's-eye-view (BEV) feature generation. Specifically, we examine the BEV feature generation method employed by the transformer-based SOTA, BEVFormer, and identify its two limitations: (i) it only generates attention weights from BEV, which precludes the use of lidar points for supervision, and (ii) it aggregates camera view features to the BEV through deformable sampling, which only selects a small subset of features and fails to exploit all information. To overcome these limitations, we propose a novel BEV feature generation method, dual-view attention, which generates attention weights from both the BEV and camera view. This method encodes all camera features into the BEV feature. By combining dual-view attention with the BEVFormer architecture, we build a new detector named VoxelFormer. Extensive experiments are conducted on the nuScenes benchmark to verify the superiority of dual-view attention and VoxelForer. We observe that even only adopting 3 encoders and 1 historical frame during training, VoxelFormer still outperforms BEVFormer significantly. When trained in the same setting, VoxelFormer can surpass BEVFormer by 4.9% NDS point. Code is available at: https://github.com/Lizhuoling/VoxelFormer-public.git.
In the field of historical manuscript research, scholars frequently encounter novel symbols in ancient texts, investing considerable effort in their identification and documentation. Although existing object detection methods achieve impressive performance on known categories, they struggle to recognize novel symbols without retraining. To address this limitation, we propose a Visually Guided Text Spotting (VGTS) approach that accurately spots novel characters using just one annotated support sample. The core of VGTS is a spatial alignment module consisting of a Dual Spatial Attention (DSA) block and a Geometric Matching (GM) block. The DSA block aims to identify, focus on, and learn discriminative spatial regions in the support and query images, mimicking the human visual spotting process. It first refines the support image by analyzing inter-channel relationships to identify critical areas, and then refines the query image by focusing on informative key points. The GM block, on the other hand, establishes the spatial correspondence between the two images, enabling accurate localization of the target character in the query image. To tackle the example imbalance problem in low-resource spotting tasks, we develop a novel torus loss function that enhances the discriminative power of the embedding space for distance metric learning. To further validate our approach, we introduce a new dataset featuring ancient Dongba hieroglyphics (DBH) associated with the Naxi minority of China. Extensive experiments on the DBH dataset and other public datasets, including EGY, VML-HD, TKH, and NC, show that VGTS consistently surpasses state-of-the-art methods. The proposed framework exhibits great potential for application in historical manuscript text spotting, enabling scholars to efficiently identify and document novel symbols with minimal annotation effort.
Autonomous driving requires an accurate and fast 3D perception system that includes 3D object detection, tracking, and segmentation. Although recent low-cost camera-based approaches have shown promising results, they are susceptible to poor illumination or bad weather conditions and have a large localization error. Hence, fusing camera with low-cost radar, which provides precise long-range measurement and operates reliably in all environments, is promising but has not yet been thoroughly investigated. In this paper, we propose Camera Radar Net (CRN), a novel camera-radar fusion framework that generates a semantically rich and spatially accurate bird's-eye-view (BEV) feature map for various tasks. To overcome the lack of spatial information in an image, we transform perspective view image features to BEV with the help of sparse but accurate radar points. We further aggregate image and radar feature maps in BEV using multi-modal deformable attention designed to tackle the spatial misalignment between inputs. CRN with real-time setting operates at 20 FPS while achieving comparable performance to LiDAR detectors on nuScenes, and even outperforms at a far distance on 100m setting. Moreover, CRN with offline setting yields 62.4% NDS, 57.5% mAP on nuScenes test set and ranks first among all camera and camera-radar 3D object detectors.
This paper shows that Masking the Deep hierarchical features is an efficient self-supervised method, denoted as MaskDeep. MaskDeep treats each patch in the representation space as an independent instance. We mask part of patches in the representation space and then utilize sparse visible patches to reconstruct high semantic image representation. The intuition of MaskDeep lies in the fact that models can reason from sparse visible patches semantic to the global semantic of the image. We further propose three designs in our framework: 1) a Hierarchical Deep-Masking module to concern the hierarchical property of patch representations, 2) a multi-group strategy to improve the efficiency without any extra computing consumption of the encoder and 3) a multi-target strategy to provide more description of the global semantic. Our MaskDeep brings decent improvements. Trained on ResNet50 with 200 epochs, MaskDeep achieves state-of-the-art results of 71.2% Top1 accuracy linear classification on ImageNet. On COCO object detection tasks, MaskDeep outperforms the self-supervised method SoCo, which specifically designed for object detection. When trained with 100 epochs, MaskDeep achieves 69.6% Top1 accuracy, which surpasses current methods trained with 200 epochs, such as HCSC, by 0.4% .
Advanced minimally invasive neurosurgery navigation relies mainly on Magnetic Resonance Imaging (MRI) guidance. MRI guidance, however, only provides pre-operative information in the majority of the cases. Once the surgery begins, the value of this guidance diminishes to some extent because of the anatomical changes due to surgery. Guidance with live image feedback coming directly from the surgical device, e.g., endoscope, can complement MRI-based navigation or be an alternative if MRI guidance is not feasible. With this motivation, we present a method for live image-only guidance leveraging a large data set of annotated neurosurgical videos.First, we report the performance of a deep learning-based object detection method, YOLO, on detecting anatomical structures in neurosurgical images. Second, we present a method for generating neurosurgical roadmaps using unsupervised embedding without assuming exact anatomical matches between patients, presence of an extensive anatomical atlas, or the need for simultaneous localization and mapping. A generated roadmap encodes the common anatomical paths taken in surgeries in the training set. At inference, the roadmap can be used to map a surgeon's current location using live image feedback on the path to provide guidance by being able to predict which structures should appear going forward or backward, much like a mapping application. Even though the embedding is not supervised by position information, we show that it is correlated to the location inside the brain and on the surgical path. We trained and evaluated the proposed method with a data set of 166 transsphenoidal adenomectomy procedures.
Domain adaptation helps generalizing object detection models to target domain data with distribution shift. It is often achieved by adapting with access to the whole target domain data. In a more realistic scenario, target distribution is often unpredictable until inference stage. This motivates us to explore adapting an object detection model at test-time, a.k.a. test-time adaptation (TTA). In this work, we approach test-time adaptive object detection (TTAOD) from two perspective. First, we adopt a self-training paradigm to generate pseudo labeled objects with an exponential moving average model. The pseudo labels are further used to supervise adapting source domain model. As self-training is prone to incorrect pseudo labels, we further incorporate aligning feature distributions at two output levels as regularizations to self-training. To validate the performance on TTAOD, we create benchmarks based on three standard object detection datasets and adapt generic TTA methods to object detection task. Extensive evaluations suggest our proposed method sets the state-of-the-art on test-time adaptive object detection task.
3D object detection is one of the most important tasks in autonomous driving and robotics. Our research focuses on tackling low efficiency issue of point-based methods on large-scale point clouds. Existing point-based methods adopt farthest point sampling (FPS) strategy for downsampling, which is computationally expensive in terms of inference time and memory consumption when the number of point cloud increases. In order to improve efficiency, we propose a novel Instance-Centroid Faster Point Sampling Module (IC-FPS) , which effectively replaces the first Set Abstraction (SA) layer that is extremely tedious. IC-FPS module is comprised of two methods, local feature diffusion based background point filter (LFDBF) and Centroid-Instance Sampling Strategy (CISS). LFDBF is constructed to exclude most invalid background points, while CISS substitutes FPS strategy by fast sampling centroids and instance points. IC-FPS module can be inserted to almost every point-based models. Extensive experiments on multiple public benchmarks have demonstrated the superiority of IC-FPS. On Waymo dataset, the proposed module significantly improves performance of baseline model and accelerates inference speed by 3.8 times. For the first time, real-time detection of point-based models in large-scale point cloud scenario is realized.
In recent years, great progress has been made in the Lift-Splat-Shot-based (LSS-based) 3D object detection method. However, inaccurate depth estimation remains an important constraint to the accuracy of camera-only and multi-model 3D object detection models, especially in regions where the depth changes significantly (i.e., the "depth jump" problem). In this paper, we proposed a novel Edge-aware Lift-splat-shot (EA-LSS) framework. Specifically, edge-aware depth fusion (EADF) module is proposed to alleviate the "depth jump" problem and fine-grained depth (FGD) module to further enforce refined supervision on depth. Our EA-LSS framework is compatible for any LSS-based 3D object detection models, and effectively boosts their performances with negligible increment of inference time. Experiments on nuScenes benchmarks demonstrate that EA-LSS is effective in either camera-only or multi-model models. It is worth mentioning that EA-LSS achieved the state-of-the-art performance on nuScenes test benchmarks with mAP and NDS of 76.5% and 77.6%, respectively.
Vision Transformers (ViTs) have been shown to be effective in various vision tasks. However, resizing them to a mobile-friendly size leads to significant performance degradation. Therefore, developing lightweight vision transformers has become a crucial area of research. This paper introduces CloFormer, a lightweight vision transformer that leverages context-aware local enhancement. CloFormer explores the relationship between globally shared weights often used in vanilla convolutional operators and token-specific context-aware weights appearing in attention, then proposes an effective and straightforward module to capture high-frequency local information. In CloFormer, we introduce AttnConv, a convolution operator in attention's style. The proposed AttnConv uses shared weights to aggregate local information and deploys carefully designed context-aware weights to enhance local features. The combination of the AttnConv and vanilla attention which uses pooling to reduce FLOPs in CloFormer enables the model to perceive high-frequency and low-frequency information. Extensive experiments were conducted in image classification, object detection, and semantic segmentation, demonstrating the superiority of CloFormer. The code is available at \url{https://github.com/qhfan/CloFormer}.
In recent years, an ever-increasing number of remote satellites are orbiting the Earth which streams vast amount of visual data to support a wide range of civil, public and military applications. One of the key information obtained from satellite imagery is to produce and update spatial maps of built environment due to its wide coverage with high resolution data. However, reconstructing spatial maps from satellite imagery is not a trivial vision task as it requires reconstructing a scene or object with high-level representation such as primitives. For the last decade, significant advancement in object detection and representation using visual data has been achieved, but the primitive-based object representation still remains as a challenging vision task. Thus, a high-quality spatial map is mainly produced through complex labour-intensive processes. In this paper, we propose a novel deep neural network, which enables to jointly detect building instance and regularize noisy building boundary shapes from a single satellite imagery. The proposed deep learning method consists of a two-stage object detection network to produce region of interest (RoI) features and a building boundary extraction network using graph models to learn geometric information of the polygon shapes. Extensive experiments show that our model can accomplish multi-tasks of object localization, recognition, semantic labelling and geometric shape extraction simultaneously. In terms of building extraction accuracy, computation efficiency and boundary regularization performance, our model outperforms the state-of-the-art baseline models.
Perception datasets for agriculture are limited both in quantity and diversity which hinders effective training of supervised learning approaches. Self-supervised learning techniques alleviate this problem, however, existing methods are not optimized for dense prediction tasks in agriculture domains which results in degraded performance. In this work, we address this limitation with our proposed Injected Noise Discriminator (INoD) which exploits principles of feature replacement and dataset discrimination for self-supervised representation learning. INoD interleaves feature maps from two disjoint datasets during their convolutional encoding and predicts the dataset affiliation of the resultant feature map as a pretext task. Our approach enables the network to learn unequivocal representations of objects seen in one dataset while observing them in conjunction with similar features from the disjoint dataset. This allows the network to reason about higher-level semantics of the entailed objects, thus improving its performance on various downstream tasks. Additionally, we introduce the novel Fraunhofer Potato 2022 dataset consisting of over 16,800 images for object detection in potato fields. Extensive evaluations of our proposed INoD pretraining strategy for the tasks of object detection, semantic segmentation, and instance segmentation on the Sugar Beets 2016 and our potato dataset demonstrate that it achieves state-of-the-art performance.
High-resolution images enable neural networks to learn richer visual representations. However, this improved performance comes at the cost of growing computational complexity, hindering their usage in latency-sensitive applications. As not all pixels are equal, skipping computations for less-important regions offers a simple and effective measure to reduce the computation. This, however, is hard to be translated into actual speedup for CNNs since it breaks the regularity of the dense convolution workload. In this paper, we introduce SparseViT that revisits activation sparsity for recent window-based vision transformers (ViTs). As window attentions are naturally batched over blocks, actual speedup with window activation pruning becomes possible: i.e., ~50% latency reduction with 60% sparsity. Different layers should be assigned with different pruning ratios due to their diverse sensitivities and computational costs. We introduce sparsity-aware adaptation and apply the evolutionary search to efficiently find the optimal layerwise sparsity configuration within the vast search space. SparseViT achieves speedups of 1.5x, 1.4x, and 1.3x compared to its dense counterpart in monocular 3D object detection, 2D instance segmentation, and 2D semantic segmentation, respectively, with negligible to no loss of accuracy.
3D object detection is an essential perception task in autonomous driving to understand the environments. The Bird's-Eye-View (BEV) representations have significantly improved the performance of 3D detectors with camera inputs on popular benchmarks. However, there still lacks a systematic understanding of the robustness of these vision-dependent BEV models, which is closely related to the safety of autonomous driving systems. In this paper, we evaluate the natural and adversarial robustness of various representative models under extensive settings, to fully understand their behaviors influenced by explicit BEV features compared with those without BEV. In addition to the classic settings, we propose a 3D consistent patch attack by applying adversarial patches in the 3D space to guarantee the spatiotemporal consistency, which is more realistic for the scenario of autonomous driving. With substantial experiments, we draw several findings: 1) BEV models tend to be more stable than previous methods under different natural conditions and common corruptions due to the expressive spatial representations; 2) BEV models are more vulnerable to adversarial noises, mainly caused by the redundant BEV features; 3) Camera-LiDAR fusion models have superior performance under different settings with multi-modal inputs, but BEV fusion model is still vulnerable to adversarial noises of both point cloud and image. These findings alert the safety issue in the applications of BEV detectors and could facilitate the development of more robust models.
Contrastive learning methods train visual encoders by comparing views from one instance to others. Typically, the views created from one instance are set as positive, while views from other instances are negative. This binary instance discrimination is studied extensively to improve feature representations in self-supervised learning. In this paper, we rethink the instance discrimination framework and find the binary instance labeling insufficient to measure correlations between different samples. For an intuitive example, given a random image instance, there may exist other images in a mini-batch whose content meanings are the same (i.e., belonging to the same category) or partially related (i.e., belonging to a similar category). How to treat the images that correlate similarly to the current image instance leaves an unexplored problem. We thus propose to support the current image by exploring other correlated instances (i.e., soft neighbors). We first carefully cultivate a candidate neighbor set, which will be further utilized to explore the highly-correlated instances. A cross-attention module is then introduced to predict the correlation score (denoted as positiveness) of other correlated instances with respect to the current one. The positiveness score quantitatively measures the positive support from each correlated instance, and is encoded into the objective for pretext training. To this end, our proposed method benefits in discriminating uncorrelated instances while absorbing correlated instances for SSL. We evaluate our soft neighbor contrastive learning method (SNCLR) on standard visual recognition benchmarks, including image classification, object detection, and instance segmentation. The state-of-the-art recognition performance shows that SNCLR is effective in improving feature representations from both ViT and CNN encoders.
Integrating LiDAR and Camera information into Bird's-Eye-View (BEV) has become an essential topic for 3D object detection in autonomous driving. Existing methods mostly adopt an independent dual-branch framework to generate LiDAR and camera BEV, then perform an adaptive modality fusion. Since point clouds provide more accurate localization and geometry information, they could serve as a reliable spatial prior to acquiring relevant semantic information from the images. Therefore, we design a LiDAR-Guided View Transformer (LGVT) to effectively obtain the camera representation in BEV space and thus benefit the whole dual-branch fusion system. LGVT takes camera BEV as the primitive semantic query, repeatedly leveraging the spatial cue of LiDAR BEV for extracting image features across multiple camera views. Moreover, we extend our framework into the temporal domain with our proposed Temporal Deformable Alignment (TDA) module, which aims to aggregate BEV features from multiple historical frames. Including these two modules, our framework dubbed BEVFusion4D achieves state-of-the-art results in 3D object detection, with 72.0% mAP and 73.5% NDS on the nuScenes validation set, and 73.3% mAP and 74.7% NDS on nuScenes test set, respectively.
Dense Self-Supervised Learning (SSL) methods address the limitations of using image-level feature representations when handling images with multiple objects. Although the dense features extracted by employing segmentation maps and bounding boxes allow networks to perform SSL for each object, we show that they suffer from coupling and positional bias, which arise from the receptive field increasing with layer depth and zero-padding. We address this by introducing three data augmentation strategies, and leveraging them in (i) a decoupling module that aims to robustify the network to variations in the object's surroundings, and (ii) a de-positioning module that encourages the network to discard positional object information. We demonstrate the benefits of our method on COCO and on a new challenging benchmark, OpenImage-MINI, for object classification, semantic segmentation, and object detection. Our extensive experiments evidence the better generalization of our method compared to the SOTA dense SSL methods