基于YOLO算法的红外图像目标检测的改进方法

传统红外图像行人检测方法利用人工进行比例模板设计和行人轮廓特征提取,由于预设模板比例相对固定,当行人因衣着增减、随身携带物品及姿态改变等原因使其轮廓比例发生较大变化时,往往会导致算法失灵而出现漏检现象。而基于深度学习的目标检测则通过对大量样本的本质特征进行抽象、提取、加工和整合,进而实现对更多样特征的学习。因此利用深度学习目标检测算法进行红外图像行人检测应用的研究可以弥补传统检测方法的不足。YOLOv3是目前性能较为均衡的识别算法,本文在分析YOLOv3系列算法的原理和特点的基础上提出了一个新的改进算法模型——Darknet-19-yolo-3,在几乎不损失检测精度的条件下提升检测速度,一定程度上实现检测准确率和速度的相对平衡。     

Wavelet based deep learning for depth estimation from single fringe pattern of fringe projection profilometry

Depth estimation from single fringe pattern is a fundamental task in the field of fringe projection three-dimensional (3D) measurement. Deep learning based on a convolutional neural network (CNN) has attracted more and more attention in fringe projection profilometry (FPP). However, most of the studies focus on complex network architecture to improve the accuracy of depth estimation with deeper and wider network architecture, which takes greater computational and lower speed. In this letter, we propose a simple method to combine wavelet transform and deep learning method for depth estimation from the single fringe pattern. Specially, the fringe pattern is decomposed into low-frequency and high-frequency details by the two-dimensional (2D) wavelet transform, which are used in the CNN network. Experiment results demonstrate that the wavelet-based deep learning method can reduce the computational complexity of the model by 4 times and improve the accuracy of depth estimation. The proposed wavelet-based deep learning models (UNet-Wavelet and hNet-Wavelet) are efficient for depth estimation of single fringe pattern, achieving better performance than the original UNet and hNet models in both qualitative and quantitative evaluation.     

基于深度卷积神经网络的红外图像行人检测

红外图像中的行人检测一直是计算机视觉领域的研究热点与难点。针对传统的红外行人检测方法需要人工设计目标表达特征的弊端,本文从深度学习的角度出发,提出一种可以自动构建目标表达特征的红外行人检测卷积神经网络。在对卷积神经网络的实现原理进行分析的基础上,设计了红外行人检测卷积神经网络的初始结构,然后通过实验对初始结构进行调整,得到最终的检测神经网络。对实拍红外人体数据库进行行人检测的实验结果表明,该方法在保持低虚警率的同时可以对红外图像中的行人进行稳健检测,优于传统方法。     

应用于嵌入式平台的实时红外行人检测方法

现有基于深度学习的远红外图像行人检测方法对计算力要求高,需要高功耗GPU计算平台,应用于嵌入式平台时,无法满足实时性和准确率需求。针对该问题,本文提出了一种新型实时红外行人检测方法,该方法使用MobileNet作为YOLOv3模型中的基础网络,辅助预测网络层以深度可分离卷积替换标准卷积,将模型改进为轻量红外行人检测模型。基于新方法构建的模型采用CVC红外行人训练集离线训练,并部署于嵌入式平台,实现红外行人在线实时检测。实验结果表明,与改进前方法相比,模型大小为65 M,约为YOLOv3的27%,新模型在基本保证原有准确率的同时,大幅降低了计算量,在同一平台下的检测速度从3FPS提升到了11FPS,可满足大部分嵌入式系统对行人检测的实时性需求。     

基于YOLO模型的红外图像行人检测方法

针对基于传统特征提取方法的远红外图像行人检测存在准确率和实时性不足的问题,本文研究了一种基于改进YOLO模型的远红外行人检测方法,通过改进其深度卷积神经网络的输入分辨率,然后在基于实际道路采集的红外数据集上进行训练,得到检测效果最佳的检测模型,并提出基于车速的自适应图像分辨率模型,以提高车载系统的行人检测性能。在基于实际道路的红外数据集上的对比实验表明,该方法与传统方法相比,准确率从76.5%提高到89.2%,每秒传输帧数从0.01259 f/s提高到40.5 f/s,满足车载情况下的实时性需求。     

Video based estimation of pedestrian walking direction for pedestrian protection system

Pedestrian protection has played an important role for driver assistance systems. Our aim is to develop a video based driver assistance system for the detection of the potentially dangerous situation between the vehicle and pedestrian, in order to warn the driver. In this paper, we address the problem of detecting pedestrian in real-world scenes and estimation of the walking direction with a single camera from a moving vehicle. Considering all the available cues for predicting the possibility of collision is very important. The direction in which the pedestrian is facing is one of the most important cues predicting where the pedestrian may move in the future. So we first address the problem of single-frame pedestrian orientation estimation in real-world scenes. Then again, we estimate the pedestrian walking direction using multi-frame based on the result of single-frame orientation estimation. We propose a three-step method: pedestrian detection for single-frame step, orientation estimation for single-frame step and walking direction estimation for multi-frame step. To evaluate the proposed method in its robustness and accuracy, the experiments have been performed between numbers of images which is highly challenging uncontrolled conditions in real world. It shows a significant performance improvement in octant orientation estimation of about 64% accuracy in the orientation estimation step and achieved surprisingly good accuracy in estimating the walking direction against 212 targeted objects.     

一种联合文本和图像信息的行人检测方法

针对纯视觉行人检测方法存在的误检、漏检率高,遮挡目标以及小尺度目标检测精度低等问题,提出一种联合文本和图像信息的行人检测方法.该方法首先利用图像分析的方法初步获取图像目标的候选框,其次通过文本分析的方法获取文本中有关图像目标的实体表达,并提出一种基于马尔科夫随机场的模型用于推断图像候选框与文本实体表达之间的共指关系（Coreference Relation）,以此达到联合图像和文本信息以辅助机器视觉提高交通场景下行人检测精度的目的.在增加了图像文本描述的加州理工大学行人检测数据集上进行的测评结果表明,该方法不仅可以在图像信息的基础上联合文本信息提高交通场景中的行人检测精度,也能在文本信息的基础上联合图像信息提高文本中的指代消解（Anaphora Resolution）精度.     

Structure-prior deep neural network for lane detection

Lane detection is an important task of road environment perception for autonomous driving. Deep learning methods based on semantic segmentation have been successfully applied to lane detection, but they require considerable computational cost for high complexity. The lane detection is treated as a particular semantic segmentation task due to the prior structural information of lane markings which have long continuous shape. Most traditional CNN are designed for the representation learning of semantic information, while this prior structural information is not fully exploited. In this paper, we propose a recurrent slice convolution module (called RSCM) to exploit the prior structural information of lane markings. The proposed RSCM is a special recurrent network structure with several slice convolution units (called SCU). The RSCM could obtain stronger semantic representation through the propagation of the prior structural information in SCU. Furthermore, we design a distance loss in consideration of the prior structure of lane markings. The lane detection network can be trained more steadily via the overall loss function formed by combining segmentation loss with the distance loss. The experimental results show the effectiveness of our method. We achieve excellent computation efficiency while keeping decent detection quality on lane detection benchmarks and the computational cost of our method is much lower than the state-of-the-art methods.     

Transformer-based global–local feature learning model for occluded person re-identification

Most recent occluded person re-identification (re-ID) methods usually learn global features directly from pedestrian images, or use additional pose estimation and semantic analysis model to learn local features, while ignoring the relationship between global and local features, thus incorrectly retrieving different pedestrians with similar attributes as the same pedestrian. Moreover, learning local features using auxiliary models brings additional computational cost. In this work, we propose a Transformer-based dual-branch feature learning model for occluded person re-ID. Firstly, we propose a global–local feature interaction module to learn the relationship between global and local features, thus enhancing the richness of information in pedestrian features. Secondly, we randomly erase local areas in the input image to simulate the real occlusion situation, thereby improving the model’s adaptability to the occlusion scene. Finally, a spilt group module is introduced to explore the local distinguishing features of pedestrian. Numerous experiments validate the effectiveness of our proposed method.     

基于特征提取与机器学习的文档区块图像分类算法

文档区块图像分类对于文档版面图像的理解和分析至关重要。在传统机器学习分类模型中,直接使用图像作为输入会导致模型参数量过大因此无法进行训练。为了克服这个困难,我们在本文中针对文档区块图像设计了一组有效的特征,并提出了基于这些特征和机器学习的文档区块分类算法。在特征设计上,我们提取了几何、灰度、区域、纹理和内容五方面在内的32维特征,以增强特征针对区块类别的分辨能力。在分类器方面,我们在所提出的特征上对传统机器学习分类模型、自动机器学习方法以及深度学习均进行了实验。在公开数据集上的实验结果表明,我们提出的文档版面区块分类算法具有很高的分类准确率,并且十分高效。另外我们实现了一个简单的分步文档版面分析算法,以展示所提出的区块分类算法的推广能力。      

基于动态加权可变形部件模型的行人检测

近年来具有代表性的工作之一是Felzenszwalb等人提出的可变形部件模型。文中从可变形部件模型中存在的一些问题展开讨论,并提出了一种基于动态加权可变形部件模型的行人检测算法。在各个部件的检测中使用动态调整权值的方法获得更加准确的判断,从而识别处于复杂环境中的行人。实验结果表明,该方法能够有效识别复杂环境中传统DPM方法难以识别的行人。     

Inexpensive fusion methods for enhancing feature detection

Recent successful approaches to high-level feature detection in image and video data have treated the problem as a pattern classification task. These typically leverage the techniques learned from statistical machine learning, coupled with ensemble architectures that create multiple feature detection models. Once created, co-occurrence between learned features can be captured to further boost performance. At multiple stages throughout these frameworks, various pieces of evidence can be fused together in order to boost performance. These approaches whilst very successful are computationally expensive, and depending on the task, require the use of significant computational resources. In this paper we propose two fusion methods that aim to combine the output of an initial basic statistical machine learning approach with a lower-quality information source, in order to gain diversity in the classified results whilst requiring only modest computing resources. Our approaches, validated experimentally on TRECVid data, are designed to be complementary to existing frameworks and can be regarded as possible replacements for the more computationally expensive combination strategies used elsewhere.     

Vehicle LED detection and segmentation recognition based on deep learning for optical camera communication

In the vehicle to vehicle (V2V) communication based on optical camera communication (OCC) system, how to achieve high reliability and low latency communication is still a problem. In this paper, we propose a lightweight light-emitting diode (LED) detection algorithm based on deep learning to detect the vehicle LED position at different communication distances, which can improve LED detection accuracy and inference speed. In addition, we design an LED segmentation recognition algorithm to reduce the bit error rate (BER) of the vehicle OCC system. The experimental results demonstrate the effectiveness of the proposed algorithms in real traffic scenes.     

基于空间结构化推理深度融合网络的RGB-D场景解析

为了弥补RGB-D场景解析中卷积神经网络空间结构化学习能力的不足,本文基于深度学习提出空间结构化推理深度融合网络,内嵌的结构化推理层有机地结合条件随机场和空间结构化推理模型,该层能够较为全面而准确地学习物体所处三维空间的物体分布以及物体间的三维空间位置关系.在此基础上,网络的特征融合层巧妙地利用深度置信网络和改进的条件随机场,该层可以根据融合生成的物体综合语义信息和物体间语义相关性信息完成深度结构化学习.实验结果表明,在标准RGB-D数据集NYUDv2和SUNRGBD上,空间结构化推理深度融合网络分别实现最优的平均准确率53.8%和54.6%,从而有助于实现机器人任务规划、车辆自动驾驶等智能计算机视觉任务.     

Orientation contrast model for boundary detection

The boundary detection task has been extensively studied in the field of computer vision and pattern recognition. Recently, researchers have formulated this task as supervised or unsupervised learning problems to leverage machine learning methods to improve detection accuracy. However, texture suppression, which is important for boundary detection, is not incorporated in this framework. To address this limitation, and also motivated by psychophysical and neurophysiological findings, we propose an orientation contrast model for boundary detection, which combines machine learning technique and texture suppression in a unified framework. Thus, the model is especially suited for detecting object boundaries surrounded by natural textures. Extensive experiments on several benchmarks demonstrate the improved boundary detection performance of the model. Specifically, its detection accuracy was improved by 10% on the Rug dataset compared with state-of-the-art unsupervised boundary detection algorithm, and its performance is also better or at least comparable with previous supervised boundary detection algorithms.     

基于可穿戴式MIMU的波峰-双阈值步数检测算法

惯性导航系统中,行人航迹推算(PDR)算法在位置解算中至关重要,其中步数统计准确程度直接影响行人的定位精度。针对传统波峰-阈值检测法存在伪波峰的影响,提出了一种基于可穿戴式微型惯性测量单元(MIMU)的波峰-双阈值步数检测算法,在行走过程中,对窗函数滤波后的合加速度进行波峰检测,并对波峰进行双阈值限定。检测到波峰满足高阈值时计为有效波峰,且相连波峰间出现低阈值时则计步成功,从而降低伪波峰对计步的影响,实现步数的精确检测。实验结果表明,当MIMU分别佩戴在多种位置时,行人多运动模式下计步精度为98%以上。     

基于强化学习的公共空间行人轨迹跟踪

对公共空间中的多目标行人轨迹跟踪问题,提出一种基于强化学习的多目标行人轨迹跟踪算法.首先采用高精确度的目标检测器检测公共空间视频中的行人目标,并为每个目标分配一个独立的单目标跟踪器进行轨迹跟踪;将每个目标作为独立智能体,通过深度强化学习方式进行训练;接下来结合跟踪轨迹与检测目标之间的表观和位置特征构建相似度代价矩阵;最...     

基于轻量化网络的光学遥感图像飞机目标检测

光学遥感图像飞机检测是遥感分析的重要研究方向。现有检测方法难以达到满意的效果,传统检测方法由于手工特征建模困难,易受背景干扰,导致其鲁棒性普遍偏低;而以复杂度提升为代价来提高检测性能的深度学习目标检测方法无法在资源受限下的星载平台得到广泛应用。针对上述问题,本论文提出一种具有轻量化多尺度特点的深度学习飞机目标检测方法。在多尺度目标检测框架（SSD）基础上,利用密集连接结构和双卷积通道构成具有特征重复利用、计算效率高等特点的基础骨干网络,之后连接一个由残差模块和反卷积构成的多尺度特征融合检测模块,以提高飞机小目标的检测性能。实验结果表明,在多种复杂机场场景中,本文的方法与当前经典的深度学习目标方法相比,在保持较高目标检测精度的同时,又能具有较低的计算复杂度。      

基于深度学习的复杂沙漠背景SAR目标检测

SAR目标检测,因成像场景大、背景复杂多变而极具挑战。传统基于恒虚警率的SAR目标检测方法极易受背景干扰。针对上述问题,提出一种基于深度学习的复杂沙漠背景SAR目标端对端检测识别系统。即采用小规模沙漠背景下的SAR图像数据对Faster-RCNN网络进行迁移训练,一体化完成典型目标的检测与识别。基于合成数据集Desert-SAR的试验结果表明,与传统方法相比,该方法检测速度更快、准确率更高、鲁棒性更强。     

Learning multiple instance deep representation for objects tracking

Object tracking has been widely used in various intelligent systems, such as pedestrian tracking, autonomous vehicles. To solve the problem that appearance changes and occlusion may lead to poor tracking performance, we propose a multiple instance learning (MIL) based method for object tracking. To achieve this task, we first manually label the first several frames of video stream in image level, which can indicate that whether a target object in the video stream. Then, we leverage a pre-trained convolutional neural network that has rich prior information to extract deep representation of target object. Since the location of the same object in adjacent frames is similar, we introduce a particle filter to predict the location of target object within a specific region. Comprehensive experiments have shown the effectiveness of our proposed method.