scholarly journals Object Detection Algorithm Based on Improved YOLOv3

Electronics ◽  
2020 ◽  
Vol 9 (3) ◽  
pp. 537 ◽  
Author(s):  
Liquan Zhao ◽  
Shuaiyang Li
Keyword(s):  
Markov Chains ◽  
Object Detection ◽  
Large Scale ◽  
Ground Truth ◽  
Detection Algorithm ◽  
Detection Methods ◽  
Cluster Method ◽  
Cluster Center ◽  
Initial Cluster ◽  
Bounding Boxes

The ‘You Only Look Once’ v3 (YOLOv3) method is among the most widely used deep learning-based object detection methods. It uses the k-means cluster method to estimate the initial width and height of the predicted bounding boxes. With this method, the estimated width and height are sensitive to the initial cluster centers, and the processing of large-scale datasets is time-consuming. In order to address these problems, a new cluster method for estimating the initial width and height of the predicted bounding boxes has been developed. Firstly, it randomly selects a couple of width and height values as one initial cluster center separate from the width and height of the ground truth boxes. Secondly, it constructs Markov chains based on the selected initial cluster and uses the final points of every Markov chain as the other initial centers. In the construction of Markov chains, the intersection-over-union method is used to compute the distance between the selected initial clusters and each candidate point, instead of the square root method. Finally, this method can be used to continually update the cluster center with each new set of width and height values, which are only a part of the data selected from the datasets. Our simulation results show that the new method has faster convergence speed for initializing the width and height of the predicted bounding boxes and that it can select more representative initial widths and heights of the predicted bounding boxes. Our proposed method achieves better performance than the YOLOv3 method in terms of recall, mean average precision, and F1-score.

scholarly journals Multi-Scale Feature Integrated Attention-Based Rotation Network for Object Detection in VHR Aerial Images

Sensors ◽  
2020 ◽  
Vol 20 (6) ◽  
pp. 1686 ◽  
Author(s):  
Feng Yang ◽  
Wentong Li ◽  
Haiwei Hu ◽  
Wanyi Li ◽  
Peng Wang
Keyword(s):  
Object Detection ◽  
Large Scale ◽  
Ground Truth ◽  
Classification Performance ◽  
Aerial Images ◽  
Detection Methods ◽  
Robust Detection ◽  
Scale Feature ◽  
Multi Scale ◽  
Bounding Boxes

Accurate and robust detection of multi-class objects in very high resolution (VHR) aerial images has been playing a significant role in many real-world applications. The traditional detection methods have made remarkable progresses with horizontal bounding boxes (HBBs) due to CNNs. However, HBB detection methods still exhibit limitations including the missed detection and the redundant detection regions, especially for densely-distributed and strip-like objects. Besides, large scale variations and diverse background also bring in many challenges. Aiming to address these problems, an effective region-based object detection framework named Multi-scale Feature Integration Attention Rotation Network (MFIAR-Net) is proposed for aerial images with oriented bounding boxes (OBBs), which promotes the integration of the inherent multi-scale pyramid features to generate a discriminative feature map. Meanwhile, the double-path feature attention network supervised by the mask information of ground truth is introduced to guide the network to focus on object regions and suppress the irrelevant noise. To boost the rotation regression and classification performance, we present a robust Rotation Detection Network, which can generate efficient OBB representation. Extensive experiments and comprehensive evaluations on two publicly available datasets demonstrate the effectiveness of the proposed framework.

scholarly journals A Lightweight Keypoint-Based Oriented Object Detection of Remote Sensing Images

2021 ◽  
Vol 13 (13) ◽  
pp. 2459
Author(s):  
Yangyang Li ◽  
Heting Mao ◽  
Ruijiao Liu ◽  
Xuan Pei ◽  
Licheng Jiao ◽  
...  
Keyword(s):  
Remote Sensing ◽  
Object Detection ◽  
Large Scale ◽  
Detection Methods ◽  
Gaussian Kernel ◽  
Remote Sensing Images ◽  
Computational Overhead ◽  
Comparable Performance ◽  
Bounding Boxes ◽  
Oriented Object

Object detection in remote sensing images has been widely used in military and civilian fields and is a challenging task due to the complex background, large-scale variation, and dense arrangement in arbitrary orientations of objects. In addition, existing object detection methods rely on the increasingly deeper network, which increases a lot of computational overhead and parameters, and is unfavorable to deployment on the edge devices. In this paper, we proposed a lightweight keypoint-based oriented object detector for remote sensing images. First, we propose a semantic transfer block (STB) when merging shallow and deep features, which reduces noise and restores the semantic information. Then, the proposed adaptive Gaussian kernel (AGK) is adapted to objects of different scales, and further improves detection performance. Finally, we propose the distillation loss associated with object detection to obtain a lightweight student network. Experiments on the HRSC2016 and UCAS-AOD datasets show that the proposed method adapts to different scale objects, obtains accurate bounding boxes, and reduces the influence of complex backgrounds. The comparison with mainstream methods proves that our method has comparable performance under lightweight.

scholarly journals A Comparative Analysis of Object Detection Metrics with a Companion Open-Source Toolkit

Electronics ◽  
2021 ◽  
Vol 10 (3) ◽  
pp. 279
Author(s):  
Rafael Padilla ◽  
Wesley L. Passos ◽  
Thadeu L. B. Dias ◽  
Sergio L. Netto ◽  
Eduardo A. B. da Silva
Keyword(s):  
Comparative Analysis ◽  
Object Detection ◽  
Open Source ◽  
Performance Metrics ◽  
Ground Truth ◽  
Detection Methods ◽  
Detection Algorithms ◽  
Spatio Temporal ◽  
Bounding Boxes ◽  
Temporal Overlap

Recent outstanding results of supervised object detection in competitions and challenges are often associated with specific metrics and datasets. The evaluation of such methods applied in different contexts have increased the demand for annotated datasets. Annotation tools represent the location and size of objects in distinct formats, leading to a lack of consensus on the representation. Such a scenario often complicates the comparison of object detection methods. This work alleviates this problem along the following lines: (i) It provides an overview of the most relevant evaluation methods used in object detection competitions, highlighting their peculiarities, differences, and advantages; (ii) it examines the most used annotation formats, showing how different implementations may influence the assessment results; and (iii) it provides a novel open-source toolkit supporting different annotation formats and 15 performance metrics, making it easy for researchers to evaluate the performance of their detection algorithms in most known datasets. In addition, this work proposes a new metric, also included in the toolkit, for evaluating object detection in videos that is based on the spatio-temporal overlap between the ground-truth and detected bounding boxes.

scholarly journals A generalised approach for high-throughput instance segmentation of stomata in microscope images

Plant Methods ◽  
2021 ◽  
Vol 17 (1) ◽  
Author(s):  
Hiranya Jayakody ◽  
Paul Petrie ◽  
Hugo Jan de Boer ◽  
Mark Whitty
Keyword(s):  
High Throughput ◽  
Imaging Techniques ◽  
Detection Algorithm ◽  
Input Image ◽  
Detection Methods ◽  
Sample Collection ◽  
High Throughput Analysis ◽  
General Applicability ◽  
Additional Image ◽  
Bounding Boxes

Abstract Background Stomata analysis using microscope imagery provides important insight into plant physiology, health and the surrounding environmental conditions. Plant scientists are now able to conduct automated high-throughput analysis of stomata in microscope data, however, existing detection methods are sensitive to the appearance of stomata in the training images, thereby limiting general applicability. In addition, existing methods only generate bounding-boxes around detected stomata, which require users to implement additional image processing steps to study stomata morphology. In this paper, we develop a fully automated, robust stomata detection algorithm which can also identify individual stomata boundaries regardless of the plant species, sample collection method, imaging technique and magnification level. Results The proposed solution consists of three stages. First, the input image is pre-processed to remove any colour space biases occurring from different sample collection and imaging techniques. Then, a Mask R-CNN is applied to estimate individual stomata boundaries. The feature pyramid network embedded in the Mask R-CNN is utilised to identify stomata at different scales. Finally, a statistical filter is implemented at the Mask R-CNN output to reduce the number of false positive generated by the network. The algorithm was tested using 16 datasets from 12 sources, containing over 60,000 stomata. For the first time in this domain, the proposed solution was tested against 7 microscope datasets never seen by the algorithm to show the generalisability of the solution. Results indicated that the proposed approach can detect stomata with a precision, recall, and F-score of 95.10%, 83.34%, and 88.61%, respectively. A separate test conducted by comparing estimated stomata boundary values with manually measured data showed that the proposed method has an IoU score of 0.70; a 7% improvement over the bounding-box approach. Conclusions The proposed method shows robust performance across multiple microscope image datasets of different quality and scale. This generalised stomata detection algorithm allows plant scientists to conduct stomata analysis whilst eliminating the need to re-label and re-train for each new dataset. The open-source code shared with this project can be directly deployed in Google Colab or any other Tensorflow environment.

scholarly journals A Set of Single YOLO Modalities to Detect Occluded Entities via Viewpoint Conversion

2021 ◽  
Vol 11 (13) ◽  
pp. 6016
Author(s):  
Jinsoo Kim ◽  
Jeongho Cho
Keyword(s):  
Object Detection ◽  
Autonomous Vehicles ◽  
Autonomous Driving ◽  
Detection Algorithm ◽  
Detection Accuracy ◽  
Cloud Data ◽  
Detection Techniques ◽  
Bounding Boxes ◽  
Partially Occluded ◽  
Rgb Image

For autonomous vehicles, it is critical to be aware of the driving environment to avoid collisions and drive safely. The recent evolution of convolutional neural networks has contributed significantly to accelerating the development of object detection techniques that enable autonomous vehicles to handle rapid changes in various driving environments. However, collisions in an autonomous driving environment can still occur due to undetected obstacles and various perception problems, particularly occlusion. Thus, we propose a robust object detection algorithm for environments in which objects are truncated or occluded by employing RGB image and light detection and ranging (LiDAR) bird’s eye view (BEV) representations. This structure combines independent detection results obtained in parallel through “you only look once” networks using an RGB image and a height map converted from the BEV representations of LiDAR’s point cloud data (PCD). The region proposal of an object is determined via non-maximum suppression, which suppresses the bounding boxes of adjacent regions. A performance evaluation of the proposed scheme was performed using the KITTI vision benchmark suite dataset. The results demonstrate the detection accuracy in the case of integration of PCD BEV representations is superior to when only an RGB camera is used. In addition, robustness is improved by significantly enhancing detection accuracy even when the target objects are partially occluded when viewed from the front, which demonstrates that the proposed algorithm outperforms the conventional RGB-based model.

Revisiting knowledge distillation for light-weight visual object detection

2021 ◽  
Vol 43 (13) ◽  
pp. 2888-2898
Author(s):  
Tianze Gao ◽  
Yunfeng Gao ◽  
Yu Li ◽  
Peiyuan Qin
Keyword(s):  
Object Detection ◽  
Essential Element ◽  
Detection Algorithm ◽  
Positive Sample ◽  
Detection Methods ◽  
Visual Object ◽  
Light Weight ◽  
Model Compression ◽  
Novel Approach ◽  
Knowledge Distillation

An essential element for intelligent perception in mechatronic and robotic systems (M&RS) is the visual object detection algorithm. With the ever-increasing advance of artificial neural networks (ANN), researchers have proposed numerous ANN-based visual object detection methods that have proven to be effective. However, networks with cumbersome structures do not befit the real-time scenarios in M&RS, necessitating the techniques of model compression. In the paper, a novel approach to training light-weight visual object detection networks is developed by revisiting knowledge distillation. Traditional knowledge distillation methods are oriented towards image classification is not compatible with object detection. Therefore, a variant of knowledge distillation is developed and adapted to a state-of-the-art keypoint-based visual detection method. Two strategies named as positive sample retaining and early distribution softening are employed to yield a natural adaption. The mutual consistency between teacher model and student model is further promoted through a hint-based distillation. By extensive controlled experiments, the proposed method is testified to be effective in enhancing the light-weight network’s performance by a large margin.

Performance comparisons of bonding box-based contact detection algorithms and a new improvement technique based on parallelization

2016 ◽  
Vol 33 (1) ◽  
pp. 7-27
Author(s):  
Mahmoud Yazdani ◽  
Hamidreza Paseh ◽  
Mostafa Sharifzadeh
Keyword(s):  
Large Scale ◽  
Distinct Element ◽  
Detection Algorithm ◽  
Computational Technique ◽  
Contact Detection ◽  
Content Type ◽  
Running Time ◽  
Spatial Sorting ◽  
Bounding Boxes ◽  
Contact Detection Algorithm

Purpose – The purpose of this paper is to find a convenient contact detection algorithm in order to apply in distinct element simulation. Design/methodology/approach – Taking the most computation effort, the performance of the contact detection algorithm highly affects the running time. The algorithms investigated in this study consist of Incremental Sort-and-Update (ISU) and Double-Ended Spatial Sorting (DESS). These algorithms are based on bounding boxes, which makes the algorithm independent of blocks shapes. ISU and DESS algorithms contain sorting and updating phases. To compare the algorithms, they were implemented in identical examples of rock engineering problems with varying parameters. Findings – The results show that the ISU algorithm gives lower running time and shows better performance when blocks are unevenly distributed in both axes. The conventional ISU merges the sorting and updating phases in its naïve implementation. In this paper, a new computational technique is proposed based on parallelization in order to effectively improve the ISU algorithm and decrease the running time of numerical analysis in large-scale rock mass projects. Originality/value – In this approach, the sorting and updating phases are separated by minor changes in the algorithm. This tends to a minimal overhead of running time and a little extra memory usage and then the parallelization of phases can be applied. On the other hand, the time consumed by the updating phase of ISU algorithm is about 30 percent of the total time, which makes the parallelization justifiable. Here, according to the results for the large-scale problems, this improved technique can increase the performance of the ISU algorithm up to 20 percent.

scholarly journals A wavelet melt detection algorithm applied to enhanced resolution scatterometer data over Antarctica (2000–2009)

2013 ◽  
Vol 7 (3) ◽  
pp. 2635-2678 ◽  
Author(s):  
N. Steiner ◽  
M. Tedesco
Keyword(s):  
Large Scale ◽  
Southern Annular Mode ◽  
Detection Algorithm ◽  
Relative Difference ◽  
Mean Value ◽  
Detection Methods ◽  
Air Temperatures ◽  
Operational Life ◽  
The Difference ◽  
Temperature Records

Abstract. Melting is mapped over Antarctica at a high spatial resolution using a novel melt-detection algorithm based on wavelets and multi-scale analysis. The method is applied to Ku band (13.4 GHz) normalized backscattering measured by SeaWinds on QuikSCAT and spatially enhanced on a 5 km grid over the operational life of the sensor (1999–2009). Wavelet-based estimates of melt spatial extent and duration are compared with those obtained by means of threshold-based detection methods, where melting is detected when the measured backscattering is 3 dB below the preceding winter mean value. Results from both methods are assessed by means of Automatic Weather Station (AWS) air surface temperature records. The yearly melting index, the product of melted area and melting duration, found using a fixed threshold and wavelet-based melt algorithm are found to have a relative difference within 7% for all years. The majority of the difference between melting records determined from QuikSCAT are related to short-duration backscatter changes identified as melting using the threshold methodology but not the wavelet-based method. Compared with AWS records both methods show a relative accuracy to within 10% based on estimated melt conditions using air temperatures. Melting maps obtained with the wavelet-based approach are also compared with those obtained from spaceborne brightness temperatures recorded by the Special Sensor Microwave/Image (SSMI). With respect to passive microwave records, we find a higher degree of agreement (9% relative difference) for the melting index using the wavelet-based approach than threshold-based methods (11% relative difference). Additionally, linkages between melting variability and the Southern Annular Mode (SAM), an important large-scale climate driver for Antarctica, are suggested by the results using wavelet based methods that are not found using threshold-based methods.

scholarly journals A Generalised Approach for High-throughput Instance Segmentation of Stomata in Microscope Images

2020 ◽  
Author(s):  
Hiranya Samanga Jayakody ◽  
Paul Petrie ◽  
Hugo de Boer ◽  
Mark Whitty
Keyword(s):  
High Throughput ◽  
Imaging Techniques ◽  
Detection Algorithm ◽  
Input Image ◽  
Detection Methods ◽  
Sample Collection ◽  
High Throughput Analysis ◽  
General Applicability ◽  
Additional Image ◽  
Bounding Boxes

Abstract Background: Stomata analysis using microscope imagery provides important insight into plant physiology, health and the surrounding environmental conditions. Plant scientists are now able to conduct automated high-throughput analysis of stomata in microscope data, however, existing detection methods are sensitive to the appearance of stomata in the training images, thereby limiting general applicability. In addition, existing methods only generate bounding-boxes around detected stomata, which require users to implement additional image processing steps to study stomata morphology. In this paper, we develop a fully automated, robust stomata detection algorithm which can also identify individual stomata boundaries regardless of the plant species, sample collection method, imaging technique and magnification level. Results: The proposed solution consists of three stages. Firstly, the input image is pre-processed to remove any colour-space biases occurring from different sample collection and imaging techniques. Secondly, a Mask R-CNN is applied to estimate individual stomata boundaries and the feature pyramid network embedded in the Mask R-CNN allows the network to identify stomata at different scales. Finally, a statistical filter is implemented at the Mask R-CNN output to reduce the number of false positive generated by the network. The algorithm was tested using 16 datasets from 12 sources, containing over 60,000 stomata. For the first time in this domain, the proposed solution was tested against 7 microscope datasets never seen by the algorithm to show the generalisability of the solution. Results indicated that the proposed approach can detect stomata with a precision, recall, and F-score of 95.10%, 83.34%, and 88.61%, respectively. A separate test conducted by comparing estimated stomata boundary values with manually measured data showed that the proposed method has an IoU score of 0.70; a 7% improvement over the bounding-box approach. Conclusion: The proposed method shows robust performance across multiple microscope image datasets of different quality and scale. This generalised stomata detection algorithm allows plant scientists to conduct stomata analysis whilst eliminating the need to re-label and re-train for each new dataset. The open-source code for the project can be directly deployed in Google Colab or any other Tensorflow environment.

scholarly journals Identify Light-curve Signals with Deep Learning Based Object Detection Algorithm. I. Transit Detection

2021 ◽  
Vol 163 (1) ◽  
pp. 23
Author(s):  
Kaiming Cui ◽  
Junjie Liu ◽  
Fabo Feng ◽  
Jifeng Liu
Keyword(s):  
Deep Learning ◽  
Object Detection ◽  
Signal To Noise Ratio ◽  
Detection Algorithm ◽  
Human Visual Perception ◽  
Lower Confidence ◽  
Learning Techniques ◽  
Confidence Threshold ◽  
Bounding Boxes ◽  
Python Package

Abstract Deep learning techniques have been well explored in the transiting exoplanet field; however, previous work mainly focuses on classification and inspection. In this work, we develop a novel detection algorithm based on a well-proven object detection framework in the computer vision field. Through training the network on the light curves of the confirmed Kepler exoplanets, our model yields about 90% precision and recall for identifying transits with signal-to-noise ratio higher than 6 (set the confidence threshold to 0.6). Giving a slightly lower confidence threshold, recall can reach higher than 95%. We also transfer the trained model to the TESS data and obtain similar performance. The results of our algorithm match the intuition of the human visual perception and make it useful to find single-transiting candidates. Moreover, the parameters of the output bounding boxes can also help to find multiplanet systems. Our network and detection functions are implemented in the Deep-Transit toolkit, which is an open-source Python package hosted on Github and PyPI.

Sign in / Sign up

Export Citation Format

Share Document