网站开发常用字体,王占山将军简介,建设人才库网站,管理咨询公司最专业的公司论文网址#xff1a;[2304.08876] 用于定向微小目标检测的动态粗到细学习 (arxiv.org)
论文代码#xff1a;https://github.com/ChaselTsui/mmrotate-dcfl
英文是纯手打的#xff01;论文原文的summarizing and paraphrasing。可能会出现难以避免的拼写错误和语法错误[2304.08876] 用于定向微小目标检测的动态粗到细学习 (arxiv.org)
论文代码https://github.com/ChaselTsui/mmrotate-dcfl
英文是纯手打的论文原文的summarizing and paraphrasing。可能会出现难以避免的拼写错误和语法错误若有发现欢迎评论指正文章偏向于笔记谨慎食用 目录
1. 省流版
1.1. 心得
1.2. 论文总结图
2. 论文逐段精读
2.1. Abstract
2.2. Introduction
2.4. Method
2.4.1. Dynamic Prior
2.4.2. Coarse Prior Matching
2.4.3. Finer Dynamic Posterior Matching
2.5. Experiments
2.5.1. Datasets
2.5.2. Implementation Details
2.5.3. Main Results
2.5.4. Ablation Study
2.6. Analysis
2.7. Conclusion
3. Reference List 1. 省流版
1.1. 心得
1为什么学脑科学的我要看这个啊愿世界上没有黑工
2最开始写小标题的时候就发现了分得好细啊好感度
3作为一个外行人这文章感觉提出了好多东西 1.2. 论文总结图 2. 论文逐段精读
2.1. Abstract ①Extreme geometric shapes (tiny) and finite features (few pixels) of tiny rotating objects will cause serious mismatch (inaccurate positional prior?) and imbalance (inaccurate positive sample features?) issues ②They proposed dynamic prior and coarse-to-fine assigner, called DCFL posterior adj.在后部的;在后面的 n.臀部;屁股 2.2. Introduction ①Oriented bounding box greatly eliminates redundant background area, especially in aerial images ②Comparison figure: where M* denotes matching function;
green, blue and red boxes are true positive, false positive, and false negative predictions respectively,
the left figure set is static and the right is dynamic ③Figure of mismatch and imbalance issues: each point in the left figure denotes a prior location先验打那么多个点啊...而且为啥打得那么整齐这是什么one-stage吗
饼状图是说当每个框都是某个角度的时候吗当每个框都不旋转的时候阳性样本平均数量是5.2还是说饼状图的意思是自由旋转某个特定角度的框的阳性样本是多少多少这个饼状图并没有横向比较诶只有这张图自己内部比较。
柱状图是锚框大小不同下平均阳性 ④They introduce dynamic Prior Capturing Block (PCB) as their prior method. Based on this, they further utilize Cross-FPN-layer Coarse Positive Sample (CPS) to assign labels. After that, they reorder these candidates by prediction (posterior), and present gt by finer Dynamic Gaussian Mixture Model (DGMM) eradicate vt.根除;消灭;杜绝 n.根除者;褪色灵 2.3. Related Work
2.3.1. Oriented Object Detection
1Prior for Oriented Objects
2Label Assignment
2.3.2. Tiny Object Detection
1Multi-scale Learning
2Label Assignment
3Context Information
4Feature Enhancement 2.4. Method
1Overview ①For a set of dense prior , where denotes width, denotes height and denotes the number of shape information什么东西啊是那些点吗, mapping it to by Deep Neural Network (DNN): where represents the detection head探测头...外行不太懂感觉也就是一个函数嘛;
one part in denotes the classification scores, where means the class number更被认为是阳性的样本那层的里的数据会更大吗;
one part in denotes the classification scores, where means the box parameter number查宝说是w, h, x, y, a之类的是box parameter ②In static methods, the pos labels assigned for is ③In dynamic methods, the pos labels set integrate posterior information: ④The loss function: where and represent the number of positive and negative samples, is the neg labels set ⑤Modelling , and : 2.4.1. Dynamic Prior ①Flexibility may alleviate mismatch problem ②Each prior represents a feature point ③The structure of Prior Capturing Block (PCB): the surrounding information is considered by dilated convolution. Then caputure dynamic prior by Deformable Convolution Network (DCN). Moreover, using the offset learned from the regression branch to guide feature extraction in the classification branch and improve alignment between the two tasks. ④To achieve dynamic prior capturing, initializing each prior loaction by each feature point’s spatial location . In each iteration, capture the offset set of each prior position to update : where denotes the stride of feature map, denotes the number of offsets;
2D Gaussian distribution is regarded as the prior distribution;
动态的作为高斯的平均向量啥玩意儿; ⑤Presetting a square on each feature point ⑥The co-variance matrix: dilate v.扩张;(使)膨胀;扩大 deformable adj.可变形的应变的易变形的 2.4.2. Coarse Prior Matching ①For prior, limiting to a single FPN may cause sub-optimal layer selection and releasing to all layers may cause slow convergence ②Therefore, they propose Cross-FPN-layer Coarse Positive Sample (CPS) candidates, expanding candidate layers to s nearby spatial location and adjacent FPN layers ③Generalized Jensen-Shannon Divergence (GJSD) constructs CPS between and : which yields a closed-form solution;
where ; and due to the homogeneity of and , ④Choosing top prior with highest GJSD for each 选差异最大的那些 2.4.3. Finer Dynamic Posterior Matching ①Two main steps are contained in this section, a posterior re-ranking strategy and a Dynamic Gaussian Mixture Model (DGMM) constraint ②The Possibility of becoming True predictions (PT) of the sample is: choosing top samples with the highest scores as Medium Positive Sample (MPS) candidates ③They apply DGMM, which contains geometry center and semantic center in one object, to filter far samples ④For specific instance , the mean vector of the first Gaussian is the geometry center , the deduced in MPS denotes semantic center ⑤Parameterizing a instance: where denotes weight of each Gaussian distribution and their summation is 1; equals to s 什么啊这是但是m可以等于1或者2诶那你g的协方差不就又是语义中心又是几何中心了吗 ⑥For any , setting negative masks 2.5. Experiments
2.5.1. Datasets ①Datasets: DOTAv1.0 /v1.5/v2.0, DIOR-R, VisDrone, and MS COCO ②Ablation dataset: DOTA-v2.0 with the most numbet of tiny objects ③Comparing dataset: DOTA-v1.0, DOTAv1.5, DOTA-v2.0, VisDrone2019, MS COCO and DIOR-R 2.5.2. Implementation Details ①Batch size: 4 ②Framework based: MMDetection and MMRotate ③Backbone: ImageNet pre-trained models ④Learning rate: 0.005 with SGD ⑤Momentum: 0.9 ⑥Weight decay: 0.0001 ⑦Default backbone: ResNet-50 with FPN ⑧Loss: Focal loss for classifying and IoU loss for regression ⑨Data augmentation: random flipping ⑩On DOTA-v1.0 and DOTA-v2.0, using official setting to crop images to 1024×1024. The overlap is 200 and epoch is 12 ⑪On other datasets, setting the input size to 1024 × 1024 (overlap 200), 800 × 800, 1333 × 800, and 1333×800 for DOTA-v1.5, DIOR-R, VisDrone, and COCO respectively. Epoch is set as 40, 40, 12, and 12 on the DOTA-v1.5, DIOR-R, COCO, and VisDrone 2.5.3. Main Results
1Results on DOTA series ①Comparison table on DOTA-v2.0 OBB: where the red ones are the best and the blue ones are the second best performance on each metric ②Comparison table on DOTA-v1.0 OBB: ③Comparison table on DOTA-v1.5 OBB: 2Results on DIOR-R ①Comparison table on DIOR-R: ②Results of typical tiny objects vehicle, bridge, and wind-mill: 3Results on HBB Datasets ①Comparison table on VisDrone, MS COCO abd DOTA-v2.0 HBB: 2.5.4. Ablation Study
1Effects of Individual Strategy ①Employ prior on each feature point ②Individual effectiveness: 2Comparisons of Different CPS ①Ablation: 3Fixed Prior and Dynamic Prior ①Ablation: 4Detailed Design in PCB ①Using the offset of the regression head to guide the offset classification head will align better than applying DCN to a single regression branch 5Effects of Parameters ①Parameter adjustment of and : ②Parameter adjustment of attenuate v. 使减弱使纤细稀薄 adj. 减弱的稀薄的细小的 2.6. Analysis
1Reconciliation of imbalance problems ①The mean predicted IoU and the mean positive sample number of holding different angles and different scales (absolute size): where the left column denotes the quality imbalance and the right column denotes the quantity imbalance. The dynamic learning from coarse to fine proposed in the paper solves the problem of sample mismatch, and more positive samples are compensated to the previous abnormal angles and scales, namely the rotated small-scale real boxes can be allocated to more positive samples than before dissection n. 解剖切开解剖体详细查究 delve vi./vt. 钻研探究挖n. 穴洞 2Visualization ①Visualization of elimilations of False Negative and False Positive predictions: where the first row and the second row are the results of RetinaNet-OBB and DCFL respectively. Furthermore, TP, FN and FP are green, red and blue frames. It can be see that DCFL can effectively locate oriented small objects with extreme shapes ②Visualization of sampled dynamic priors: 3Speed ①Compared with R3Det, S 2A-Net and RetinaNet with 16.2, 18.9, 20.8, FPS of DCFL is 20.9, which means the high efficiency of DCFL ②Parameters and GLOPs of DCFL: 2.7. Conclusion For solving the problems of mismatched feature prior and unbalanced positive samples, the authors proposed DCFL model with dynamic prior and coarse-to-fine assigner. Ultimately,, it achieves a remarkable performance 3. Reference List
Xu, C. et al. (2023) Dynamic Coarse-to-Fine Learning for Oriented Tiny Object Detection, CVPR. doi: https://doi.org/10.48550/arXiv.2304.08876
