Faster Region-based Convolutional Network Research Articles

Black-box adversarial patch attacks using differential evolution against aerial imagery object detectors

Adversarial patch attacks are commonly used to attack aerial imagery object detectors. However, most existing methods are designed for white-box settings, which are impractical in real world scenarios. In this work, we propose a novel framework for black-box adversarial patch attacks against aerial imagery object detectors using differential evolution (DE). Specifically, we first introduce a new dimensionality reduction strategy to address the high-dimensionality curse in DE optimization problems. Then, we design three universal fitness functions to help DE to select better individuals within a limited computational budget. Finally, we conduct extensive experiments on the Dataset for Object Detection in Aerial images (DOTA) against You Only Look Once (YOLOv3, YOLOv4), and Faster Region-based Convolutional Network (Faster R-CNN), utilizing attack success rate (ASR) and average precision (AP) to quantitatively evaluate the attack efficiency. Results on YOLOv3 for the plane category indicate that our proposed strategy achieves at least a 20.97% improvement in ASR compared to existing dimensionality reduction schemes. Moreover, experiments conducted on detectors with different architectures reveal inconsistent adversarial robustness across categories. To the best of our knowledge, this is the first work to explore the use of DE in black-box adversarial patch attacks against aerial imagery object detectors. Our code has been released at https://github.com/tang-agui/Bbox-Att-Detector.

Comparative Evaluation of Convolutional Neural Network Object Detection Algorithms for Vehicle Detection.

The domain of object detection was revolutionized with the introduction of Convolutional Neural Networks (CNNs) in the field of computer vision. This article aims to explore the architectural intricacies, methodological differences, and performance characteristics of three CNN-based object detection algorithms, namely Faster Region-Based Convolutional Network (R-CNN), You Only Look Once v3 (YOLO), and Single Shot MultiBox Detector (SSD) in the specific domain application of vehicle detection. The findings of this study indicate that the SSD object detection algorithm outperforms the other approaches in terms of both performance and processing speeds. The Faster R-CNN approach detected objects in images with an average speed of 5.1 s, achieving a mean average precision of 0.76 and an average loss of 0.467. YOLO v3 detected objects with an average speed of 1.16 s, achieving a mean average precision of 0.81 with an average loss of 1.183. In contrast, SSD detected objects with an average speed of 0.5 s, exhibiting the highest mean average precision of 0.92 despite having a higher average loss of 2.625. Notably, all three object detectors achieved an accuracy exceeding 99%.

Lightweight Detection Methods for Insulator Self-Explosion Defects.

The accurate and efficient detection of defective insulators is an essential prerequisite for ensuring the safety of the power grid in the new generation of intelligent electrical system inspections. Currently, traditional object detection algorithms for detecting defective insulators in images face issues such as excessive parameter size, low accuracy, and slow detection speed. To address the aforementioned issues, this article proposes an insulator defect detection model based on the lightweight Faster R-CNN (Faster Region-based Convolutional Network) model (Faster R-CNN-tiny). First, the Faster R-CNN model's backbone network is turned into a lightweight version of it by substituting EfficientNet for ResNet (Residual Network), greatly decreasing the model parameters while increasing its detection accuracy. The second step is to employ a feature pyramid to build feature maps with various resolutions for feature fusion, which enables the detection of objects at various scales. In addition, replacing ordinary convolutions in the network model with more efficient depth-wise separable convolutions increases detection speed while slightly reducing network detection accuracy. Transfer learning is introduced, and a training method involving freezing and unfreezing the model is employed to enhance the network's ability to detect small target defects. The proposed model is validated using the insulator self-exploding defect dataset. The experimental results show that Faster R-CNN-tiny significantly outperforms the Faster R-CNN (ResNet) model in terms of mean average precision (mAP), frames per second (FPS), and number of parameters.

Bi-LS-AttM: A Bidirectional LSTM and Attention Mechanism Model for Improving Image Captioning

The discipline of automatic image captioning represents an integration of two pivotal branches of artificial intelligence, namely computer vision (CV) and natural language processing (NLP). The principal functionality of this technology lies in transmuting the extracted visual features into semantic information of a higher order. The bidirectional long short-term memory (Bi-LSTM) has garnered wide acceptance in executing image captioning tasks. Of late, scholarly attention has been focused on modifying suitable models for innovative and precise subtitle captions, although tuning the parameters of the model does not invariably yield optimal outcomes. Given this, the current research proposes a model that effectively employs the bidirectional LSTM and attention mechanism (Bi-LS-AttM) for image captioning endeavors. This model exploits the contextual comprehension from both anterior and posterior aspects of the input data, synergistically with the attention mechanism, thereby augmenting the precision of visual language interpretation. The distinctiveness of this research is embodied in its incorporation of Bi-LSTM and the attention mechanism to engender sentences that are both structurally innovative and accurately reflective of the image content. To enhance temporal efficiency and accuracy, this study substitutes convolutional neural networks (CNNs) with fast region-based convolutional networks (Fast RCNNs). Additionally, it refines the process of generation and evaluation of common space, thus fostering improved efficiency. Our model was tested for its performance on Flickr30k and MSCOCO datasets (80 object categories). Comparative analyses of performance metrics reveal that our model, leveraging the Bi-LS-AttM, surpasses unidirectional and Bi-LSTM models. When applied to caption generation and image-sentence retrieval tasks, our model manifests time economies of approximately 36.5% and 26.3% vis-a-vis the Bi-LSTM model and the deep Bi-LSTM model, respectively.

A quantitative heterogeneity analysis approach to molecular subtype recognition of breast cancer in dynamic contrast-enhanced magnetic imaging images from radiomics data.

Breast cancer is a major cause of mortality among women worldwide. Dynamic contrast-enhanced breast magnetic resonance imaging (DCE-MRI) is a good imaging technique that can show temporal information about the kinetics of the contrast agent in suspicious breast lesions as well as acceptable spatial resolution. Computer-aided detection systems assist in the detection of lesions through medical image processing techniques combined with computerized analysis and calculation, which in turn helps radiologists recognize molecular subtypes of breast lesions that will be beneficial for better treatment plan decisions. In this paper, a computer-aided diagnosis method is proposed to automatically locate breast cancer lesions and identify molecular subtypes of breast cancer with heterogeneity analysis from radiomics data. A fast region-based convolutional network (Faster R-CNN) framework is first applied to images to detect breast cancer lesions. Then, the heterogeneous regions of every breast cancer lesion are extracted. Based on the multiple visual and kinetic radiomics features extracted from the heterogeneous regions, a temporal bag of visual word model is proposed, which takes into account the dynamic characteristics of both lesion and heterogeneous regions in images over time. The recognition task of molecular subtypes of breast lesions is realized based on a stacking classification model. At the genetic level, breast cancer is divided into four molecular subtypes, namely, luminal epithelial type A (Luminal A), luminal epithelial type B (Luminal B), HER-2 overexpression and basal cell type. The experimental results show that the precision of the four subtypes is 93%, 94%, 83%, 86%; the recall is 96%, 80%, 91%, 94%; and the F1-score is 95%, 86%, 87%. The experimental results denote the influence of heterogeneous regions on the recognition task. The DCE-MRI-based approach to identify molecular typing of breast cancer for noninvasive diagnosis will contribute to the development of breast cancer treatment, improved outcomes and reduced mortality.

AI-assisted identification of intrapapillary capillary loops in magnification endoscopy for diagnosing early-stage esophageal squamous cell carcinoma: a preliminary study.

Esophageal squamous cell carcinoma (ESCC) is one of the most common histological types of esophageal cancers. It can seriously affect public health, particularly in Eastern Asia. Early diagnosis and effective therapy of ESCC can significantly help improve patient prognoses. The visualization of intrapapillary capillary loops (IPCLs) under magnification endoscopy (ME) can greatly support the identification of ESCC occurrences by endoscopists. This paper proposes an artificial-intelligence-assisted endoscopic diagnosis approach using deep learning for localizing and identifying IPCLs to diagnose early-stage ESCC. An improved Faster region-based convolutional network (R-CNN) with a polarized self-attention (PSA)-HRNetV2p backbone was employed to automatically detect IPCLs in ME images. In our study, 2887 ME with blue laser imaging (ME-BLI) images of 246 patients and 493 ME with narrow-band imaging (ME-NBI) images of 81 patients were collected from multiple hospitals and used to train and test our detection model. The ME-NBI images were used as the external testing set to verify the generalizability of the model. The experimental evaluation revealed that the proposed method achieved a recall of 79.25%, precision of 75.54%, F1-score of 0.764 and mean average precision (mAP) of 74.95%. Our method outperformed other existing approaches in our evaluation. It can effectively improve the accuracy of ESCC detection and provide a useful adjunct to the assessment of early-stage ESCC for endoscopists.

Tooth-Related Disease Detection System Based on Panoramic Images and Optimization Through Automation: Development Study.

Early detection of tooth-related diseases in patients plays a key role in maintaining their dental health and preventing future complications. Since dentists are not overly attentive to tooth-related diseases that may be difficult to judge visually, many patients miss timely treatment. The 5 representative tooth-related diseases, that is, coronal caries or defect, proximal caries, cervical caries or abrasion, periapical radiolucency, and residual root can be detected on panoramic images. In this study, a web service was constructed for the detection of these diseases on panoramic images in real time, which helped shorten the treatment planning time and reduce the probability of misdiagnosis. This study designed a model to assess tooth-related diseases in panoramic images by using artificial intelligence in real time. This model can perform an auxiliary role in the diagnosis of tooth-related diseases by dentists and reduce the treatment planning time spent through telemedicine. For learning the 5 tooth-related diseases, 10,000 panoramic images were modeled: 4206 coronal caries or defects, 4478 proximal caries, 6920 cervical caries or abrasion, 8290 periapical radiolucencies, and 1446 residual roots. To learn the model, the fast region-based convolutional network (Fast R-CNN), residual neural network (ResNet), and inception models were used. Learning about the 5 tooth-related diseases completely did not provide accurate information on the diseases because of indistinct features present in the panoramic pictures. Therefore, 1 detection model was applied to each tooth-related disease, and the models for each of the diseases were integrated to increase accuracy. The Fast R-CNN model showed the highest accuracy, with an accuracy of over 90%, in diagnosing the 5 tooth-related diseases. Thus, Fast R-CNN was selected as the final judgment model as it facilitated the real-time diagnosis of dental diseases that are difficult to judge visually from radiographs and images, thereby assisting the dentists in their treatment plans. The Fast R-CNN model showed the highest accuracy in the real-time diagnosis of dental diseases and can therefore play an auxiliary role in shortening the treatment planning time after the dentists diagnose the tooth-related disease. In addition, by updating the captured panoramic images of patients on the web service developed in this study, we are looking forward to increasing the accuracy of diagnosing these 5 tooth-related diseases. The dental diagnosis system in this study takes 2 minutes for diagnosing 5 diseases in 1 panoramic image. Therefore, this system plays an effective role in setting a dental treatment schedule.

Pedestrian detection system based on deep learning

<p>Pedestrian detection is a rapidly growing field of computer vision with applications in smart cars, surveillance, automotive safety, and advanced robotics. Most of the success of the last few years has been driven by the rapid growth of deep learning, more efficient tools capable of learning semantic, high-level, deeper features of images are proposed. In this article, we investigated the task of pedestrian detection on roads using models based on convolutional neural networks. We compared the performance of standard state-of-the-art object detectors like Faster region-based convolutional network (R-CNN), single shot detector (SSD), and you only look once, version 3 (YOLOv3). Results show that YOLOv3 is the best object detection model than others for pedestrians in terms of detection and time prediction.</p>

Tracking perovskite crystallization via deep learning-based feature detection on 2D X-ray scattering data

Understanding the processes of perovskite crystallization is essential for improving the properties of organic solar cells. In situ real-time grazing-incidence X-ray diffraction (GIXD) is a key technique for this task, but it produces large amounts of data, frequently exceeding the capabilities of traditional data processing methods. We propose an automated pipeline for the analysis of GIXD images, based on the Faster Region-based Convolutional Network architecture for object detection, modified to conform to the specifics of the scattering data. The model exhibits high accuracy in detecting diffraction features on noisy patterns with various experimental artifacts. We demonstrate our method on real-time tracking of organic-inorganic perovskite structure crystallization and test it on two applications: 1. the automated phase identification and unit-cell determination of two coexisting phases of Ruddlesden–Popper 2D perovskites, and 2. the fast tracking of MAPbI3 perovskite formation. By design, our approach is equally suitable for other crystalline thin-film materials.

Smoke Detection Algorithm based on Negative Sample Mining

Forest fire is one of the most dangerous disasters that threaten the safety of human life and property. In order to detect fire in time, we detect the smoke when the fire breaks out. However, it is still a challenging task due to the variations of smoke in color, texture, shape and the disturbances of smoke-like objects. Therefore, the accuracy of smoke detection is not high, and it is accompanied by a high false positive rate, especially in the real environment. To tackle this problem, this paper proposes a novel model based on Faster Region-based Convolutional Network (R-CNN) which utilizes negative sample mining method. The proposed method allows the model to learn more negative sample features, thereby reducing false positives in smoke detection. The experiments are performed on self-created dataset containing 11958 images which are collected from cameras placed in villages or towns and existing datasets. Compared to other smoke datasets, the self-created dataset is larger and contains complex scenes. The proposed method achieves 94.59% accuracy, 94.35% precision and 5.76% false positive rate on self-created dataset. The results show that the proposed network is better and more robust than previous works.

Special Convolutional Neural Network for Identification and Positioning of Interstitial Lung Disease Patterns in Computed Tomography Images

In this paper, automated detection of interstitial lung disease patterns in high resolution computed tomography images is achieved by developing a faster region-based convolutional network based detector with GoogLeNet as a backbone. GoogLeNet is simplified by removing few inception models and used as the backbone of the detector network. The proposed framework is developed to detect several interstitial lung disease patterns without doing lung field segmentation. The proposed method is able to detect the five most prevalent interstitial lung disease patterns: fibrosis, emphysema, consolidation, micronodules and ground-glass opacity, as well as normal. Five-fold cross-validation has been used to avoid bias and reduce over-fitting. The proposed framework performance is measured in terms of F-score on the publicly available MedGIFT database. It outperforms state-of-the-art techniques. The detection is performed at slice level and could be used for screening and differential diagnosis of interstitial lung disease patterns using high resolution computed tomography images.

Scene recognition of road traffic accident based on an improved faster R-CNN algorithm

Traffic accident scene recognition is the first premise of accident analysis and scene reconstruction, so the research on traffic accident scene recognition technology is of great significance. In this paper, an improved Region Proposal Network based on Faster Region-based Convolutional Network (Faster R-CNN) is proposed. Because the network adopts four strategies to improve the Regional Proposal Network (RPN), the network structure in this paper is called Multi-strategy Region Proposal Network (MSRPN). The experimental results show that the mAP value of MSRPN algorithm surpasses the other two target recognition algorithms. At the same time, MSRPN only needs to generate 150 region proposals in each image to obtain the above experimental results. In addition, the algorithm has good performance in small target detection. Especially, the target recognition speed is 6 fps, which is faster than other target detection algorithms. In conclusion, the target recognition algorithm based on MSRPN can be effectively applied to target recognition system.

Computer Vision Based Automatic Recognition of Pointer Instruments: Data Set Optimization and Reading.

With the promotion of intelligent substations, more and more robots have been used in industrial sites. However, most of the meter reading methods are interfered with by the complex background environment, which makes it difficult to extract the meter area and pointer centerline, which is difficult to meet the actual needs of the substation. To solve the current problems of pointer meter reading for industrial use, this paper studies the automatic reading method of pointer instruments by putting forward the Faster Region-based Convolutional Network (Faster-RCNN) based object detection integrating with traditional computer vision. Firstly, the Faster-RCNN is used to detect the target instrument panel region. At the same time, the Poisson fusion method is proposed to expand the data set. The K-fold verification algorithm is used to optimize the quality of the data set, which solves the lack of quantity and low quality of the data set, and the accuracy of target detection is improved. Then, through some image processing methods, the image is preprocessed. Finally, the position of the centerline of the pointer is detected by the Hough transform, and the reading can be obtained. The evaluation of the algorithm performance shows that the method proposed in this paper is suitable for automatic reading of pointer meters in the substation environment, and provides a feasible idea for the target detection and reading of pointer meters.

Application of conventional UAV-based high-throughput object detection to the early diagnosis of pine wilt disease by deep learning

The early diagnosis of pine wilt disease (PWD) is crucial to its management. Substantial effort has been made to develop an accurate early diagnosis method. However, none of the existing methods are suitable for large-scale rapid screening in the field. In this study, an unmanned aerial vehicle (UAV) was used to collect a large number of images from a pine tree canopy in an early stage of infection for the generation of a training dataset. Owing to the inability to develop object detection models using nine regular machine learning classifiers, two advanced deep learning algorithms were employed, namely the Faster Region-based Convolutional Network (Faster R-CNN) and You Only Look Once version 3 (YOLOv3). Model performances were compared based on the precision (mAP), size, and processing speed. All four models possessed similar precision (0.602–0.64), but the YOLO-based models had a smaller size and faster processing speed than the Faster R-CNN-adapted models. The population of infected trees (in the early or late stage of infection) was predicted under different treatments (retaining or removing the dead trees from the forest) using these four models to explore their application. The results indicate that retaining dead trees after chopping them down results in fewer dead trees but more early infected trees the following year. We propose a cost-effective and high-throughput method for the early diagnosis of PWD in the field using UAV-based image processing and object detection (uses YOLOv3) based on deep learning.

Faster R-CNN Learning-Based Semantic Filter for Geometry Estimation and Its Application in vSLAM Systems

Epipolar geometry is a fundamental constraint used in computer vision systems to estimate parameters using correspondence. The most common way to describe epipolar geometry is by means of a <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$3\times 3$ </tex-math></inline-formula> matrix called the fundamental matrix, and such matrices are used to store the precise geometric information relating a pair of stereo images. Its efficient estimation substantially improves initialization in visual simultaneous localization and mapping (vSLAM), which uses correspondence-based epipolar geometry to determine the trajectory of the camera and a three-dimensional scene. Conventional robust methods for epipolar geometry estimation can become computationally inefficient and inaccurate when there are low-quality correspondences. Because semantic information can be more stable than pixel intensities/descriptors, a novel Faster Region-based Convolutional Network (R-CNN) learning-based approach called the semantic filter is proposed in this paper to address these problems. The semantic filter is first trained on different semantic patches, which are described in terms of their different outlier distributions, providing different semantic labels for image contexts. Then, the patches with low-level semantic labels are filtered out. Finally, precise and robust correspondences can be determined by matches using the high-level semantic contexts, making the correspondence-based calculation more accurate. For dynamic outdoor scenes, the results of extensive experiments show that our semantic filter can help vSLAM localize accurately and robustly on a map from different viewpoints. In a completely static scenario, our semantic filter can remove the low-quality correspondences, enabling the mobile robot to operate well.

Reliable Thyroid Carcinoma Detection with Real-Time Intelligent Analysis of Ultrasound Images

Thyroid carcinoma is one of the most common endocrine diseases globally, and the incidence has been on the rise in recent years. Ultrasound imaging is the primary clinical method for early thyroid nodule diagnosis. Regions of interest (ROIs) of nodules in ultrasound images are difficult to detect because of their irregular shape nand vague margins. Accurate real-time thyroid nodule detection can provide ROIs for subsequent nodule diagnosis automatically, avoid variabilities between the subjective interpretations and inter-observer effectively and alleviate the workloads of medical practitioners. The aim of this study was to present a reliable, real-time detection method based on the Faster R-CNN (region-based convolutional network) framework for accurate and fast detection of thyroid nodules in ultrasound images. Our study proposed a faster and more accurate thyroid nodule detection method based on the Faster R-CNN framework by adding three strategies: feature pyramid, spatial remapping and anchor-box redesign. Specifically, the network takes raw ultrasound images as inputs and generates boxes with positions and the possibilities that these boxes contain thyroid nodules. The proposed method could locate and detect target nodules accurately with a mean average precision of 92.79% with more than 9000 patient images. In addition, the detection rate has accelerated to >16 frames per second, four times faster than that of the initial network. Therefore, it can meet the requirements of clinical application. The performance of the fivefold cross-validation was also accurate and robust. The proposed automatic thyroid nodule detection method yields better performance in accuracy and detection speed, which indicates the potential value of our method in assisting clinical ultrasound image interpretation.

Computer-Aided System for the Detection of Multicategory Pulmonary Tuberculosis in Radiographs.

The early screening and diagnosis of tuberculosis plays an important role in the control and treatment of tuberculosis infections. In this paper, an integrated computer-aided system based on deep learning is proposed for the detection of multiple categories of tuberculosis lesions in chest radiographs. In this system, the fully convolutional neural network method is used to segment the lung area from the entire chest radiograph for pulmonary tuberculosis detection. Different from the previous analysis of the whole chest radiograph, we focus on the specific tuberculosis lesion areas for the analysis and propose the first multicategory tuberculosis lesion detection method. In it, a learning scalable pyramid structure is introduced into the Faster Region-based Convolutional Network (Faster RCNN), which effectively improves the detection of small-area lesions, mines indistinguishable samples during the training process, and uses reinforcement learning to reduce the detection of false-positive lesions. To compare our method with the current tuberculosis detection system, we propose a classification rule for whole chest X-rays using a multicategory tuberculosis lesion detection model and achieve good performance on two public datasets (Montgomery: AUC = 0.977 and accuracy = 0.926; Shenzhen: AUC = 0.941 and accuracy = 0.902). Our proposed computer-aided system is superior to current systems that can be used to assist radiologists in diagnoses and public health providers in screening for tuberculosis in areas where tuberculosis is endemic.

Arbitrary Direction Ship Detection in Remote-Sensing Images Based on Multitask Learning and Multiregion Feature Fusion

Ship detection in remote sensing plays an important role in civil and military fields. Owing to the complex background and uncertain direction, ship detection is full of challenge by using the commonly used object-detection methods. In this article, a new framework for detecting the arbitrary direction ships is proposed based on the improvement in the Faster region-based convolutional network (R-CNN), in which the shape of the bounding box is described by three sides, namely, vertical side, horizontal side, and short side, respectively. The inclination of the ship is obtained by calculating the arc-tangent value of the vertical side to the horizontal side. First, the better performing ResNet-101 is adopted to extract features over an entire image, which are shared by the region proposal network (RPN) and the head network. Then, the multidirection proposal regions that may contain ships are generated by the RPN. Next, the global and local features of the proposal regions are combined as the whole features of the regions by a multiregion feature-fusion (MFF) module, which can provide more detailed information of the regions. Finally, the head network uses the whole features of the proposal regions for bounding-box recognition through multitask learning, including classification, regression, and incline direction prediction (left or right). The proposed method is tested and compared with other state-of-the-art ship-detection methods on two open remote-sensing data sets and some large-scale and real images. The experimental results validate that the proposed approach has achieved better performance.

Small-Object Detection in Remote Sensing Images with End-to-End Edge-Enhanced GAN and Object Detector Network

The detection performance of small objects in remote sensing images has not been satisfactory compared to large objects, especially in low-resolution and noisy images. A generative adversarial network (GAN)-based model called enhanced super-resolution GAN (ESRGAN) showed remarkable image enhancement performance, but reconstructed images usually miss high-frequency edge information. Therefore, object detection performance showed degradation for small objects on recovered noisy and low-resolution remote sensing images. Inspired by the success of edge enhanced GAN (EEGAN) and ESRGAN, we applied a new edge-enhanced super-resolution GAN (EESRGAN) to improve the quality of remote sensing images and used different detector networks in an end-to-end manner where detector loss was backpropagated into the EESRGAN to improve the detection performance. We proposed an architecture with three components: ESRGAN, EEN, and Detection network. We used residual-in-residual dense blocks (RRDB) for both the ESRGAN and EEN, and for the detector network, we used a faster region-based convolutional network (FRCNN) (two-stage detector) and a single-shot multibox detector (SSD) (one stage detector). Extensive experiments on a public (car overhead with context) dataset and another self-assembled (oil and gas storage tank) satellite dataset showed superior performance of our method compared to the standalone state-of-the-art object detectors.

Bending diagnosis of rice seedling lines and guidance line extraction of automatic weeding equipment in paddy field

Mechanical weeding is an efficient weeding method, which is of considerable significance to the paddy field ecosystem. However, traditional mechanical weeding methods can cause seedling damages due to the bending phenomenon of the seedling lines. Introducing computer vision and control technology to traditional mechanical weeding methods can help the system diagnose the bending phenomenon and avoid crushing the seedlings. In this paper, we propose a deep-learning-based method of seedling line bending diagnosis and guidance line extraction. To prove the proposed method effective in the mechanical weeding system, we choose the Faster Region-based Convolutional Network (R-CNN) and Single Shot MultiBox Detector (SSD) as the representative models of the single-phase method and the two-phase method. With a novel dataset of rice seedling images established, we compare and analyze the confidence and real-time performance of the trained models. The experimental results show that the Faster R-CNN model is better in terms of accuracy, yet the SSD model has more advantages in the speed. Comprehensively considering the system requiring and model performances, the SSD model is a better choice in the automatic rice avoidance system.