In the past several years,remarkable achievements have been made in the field of object detection.Although performance is generally improving,the accuracy of small object detection remains low compared with that of la...In the past several years,remarkable achievements have been made in the field of object detection.Although performance is generally improving,the accuracy of small object detection remains low compared with that of large object detection.In addition,localization misalignment issues are common for small objects,as seen in GoogLeNets and residual networks(ResNets).To address this problem,we propose an improved region-based fully convolutional network(R-FCN).The presented technique improves detection accuracy and eliminates localization misalignment by replacing positionsensitive region of interest(PS-RoI)pooling with position-sensitive precise region of interest(PS-Pr-RoI)pooling,which avoids coordinate quantization and directly calculates two-order integrals for position-sensitive score maps,thus preventing a loss of spatial precision.A validation experiment was conducted in which the Microsoft common objects in context(MS COCO)training dataset was oversampled.Results showed an accuracy improvement of 3.7%for object detection tasks and an increase of 6.0%for small objects.展开更多
Spike development directly affects the yield and quality of rice. We describe an algorithm for automatically identifying multiple developmental stages of rice spikes(AI-MDSRS) that transforms the automatic identificat...Spike development directly affects the yield and quality of rice. We describe an algorithm for automatically identifying multiple developmental stages of rice spikes(AI-MDSRS) that transforms the automatic identification of multiple developmental stages of rice spikes into the detection of rice spikes of diverse maturity levels. The scales vary greatly in different growth and development stages because rice spikes are dense and small, posing challenges for their effective and accurate detection. We describe a rice spike detection model based on an improved faster regions with convolutional neural network(Faster R-CNN).The model incorporates the following optimization strategies: first, Inception_Res Net-v2 replaces VGG16 as a feature extraction network;second, a feature pyramid network(FPN) replaces single-scale feature maps to fuse with region proposal network(RPN);third, region of interest(Ro I) alignment replaces Ro I pooling, and distance-intersection over union(DIo U) is used as a standard for non-maximum suppression(NMS). The performance of the proposed model was compared with that of the original Faster R-CNN and YOLOv4 models. The mean average precision(m AP) of the rice spike detection model was92.47%, a substantial improvement on the original Faster R-CNN model(with 40.96% m AP) and 3.4%higher than that of the YOLOv4 model, experimentally indicating that the model is more accurate and reliable. The identification results of the model for the heading–flowering, milky maturity, and full maturity stages were within two days of the results of manual observation, fully meeting the needs of agricultural activities.展开更多
Background： The role of rituximab in combination with CHOP regimen in patients with stage I diffuse large B-cell lymphoma （DLBCL） remains to be defined. We aimed to compare CHOP plus rituximab （R-CHOP） with CHOP ...Background： The role of rituximab in combination with CHOP regimen in patients with stage I diffuse large B-cell lymphoma （DLBCL） remains to be defined. We aimed to compare CHOP plus rituximab （R-CHOP） with CHOP alone and determine the value of radiotherapy in these patients. Methods： Between 2003 and 2009, 140 untreated patients with stage I DLBCL were retrospectively analyzed in this study. Results： Seventy-eight patients were treated in R-CHOP group and 62 in CHOP group. Ninety-one patients received additional radiotherapy at the end of chemotherapy. The different treatment groups were well-balanced with respect to baseline characteristics. Complete response （CR） rate was 77% both in R-CHOP and CHOP groups （P=0.945）. After a median follow-up period of 56 months, patients received R-CHOP regimen had similar 5-year progression-free survival （PFS） （76% vs. 85%; log-rank P=0.215） and 5-year overall survival （OS） （90% vs. 96%; log-rank P=0.175） compared with those with CHOP alone. Patients with radiotherapy had significantly increased 5-year PFS compared with those who had chemotherapy alone （86% vs. 71%; log-rank P=0.005）. At multivariate analysis, patients who had CR （P=0.008） and received radiotherapy （P=0.003） were significantly associated with superior PFS. Conclusions： CHOP alone could be as effective as R-CHOP regimen and additional radiotherapy would be necessary for stage I or stage I non-bulky DLBCL patients.展开更多
After defining the strong tensor product of strong (sub)chain complenes, it is shown that an analogue of the Kunneth theorem holds in strong homology by proving that the kernel (cokernel) of connecting homomorphisms i...After defining the strong tensor product of strong (sub)chain complenes, it is shown that an analogue of the Kunneth theorem holds in strong homology by proving that the kernel (cokernel) of connecting homomorphisms is isomorphic to the direct sum of torsion (tensor) products of strong homology groups. An isomorphism between strong (r-stage) homology groups of inverse systems is also constructed.展开更多
Data envelopment analysis(DEA)is a technique to measure the performance of decision-making units(DMUs).Conventional DEA treats DMUs as black boxes and the internal structure of DMUs is ignored.Two-stage DEA models are...Data envelopment analysis(DEA)is a technique to measure the performance of decision-making units(DMUs).Conventional DEA treats DMUs as black boxes and the internal structure of DMUs is ignored.Two-stage DEA models are special case network DEA models that explore the internal structures of DMUs.Most often,one output cannot be produced by certain input data and/or the data may be expressed as ratio output/input.In these cases,traditional two-stage DEA models can no longer be used.To deal with these situations,we applied DEA-Ratio(DEA-R)to evaluate two-stage DMUs instead of traditional DEA.To this end,we developed two novel DEA-R models,namely,range directional DEA-R(RDD-R)and(weighted)Tchebycheff norm DEA-R(TND-R).The validity and reliability of our proposed approaches are shown by some examples.The Taiwan Residents non-life insurance companies are revisited using these proposed approaches and the results from the proposed methods are compared with those from some other methods.展开更多
基金This project was supported by the National Natural Science Foundation of China under grant U1836208the Hunan Provincial Natural Science Foundations of China under Grant 2020JJ4626+2 种基金the Scientific Research Fund of Hunan Provincial Education Department of China under Grant 19B004the“Double First-class”International Cooperation and Development Scientific Research Project of Changsha University of Science and Technology under Grant 2018IC25the Young Teacher Growth Plan Project of Changsha University of Science and Technology under Grant 2019QJCZ076.
文摘In the past several years,remarkable achievements have been made in the field of object detection.Although performance is generally improving,the accuracy of small object detection remains low compared with that of large object detection.In addition,localization misalignment issues are common for small objects,as seen in GoogLeNets and residual networks(ResNets).To address this problem,we propose an improved region-based fully convolutional network(R-FCN).The presented technique improves detection accuracy and eliminates localization misalignment by replacing positionsensitive region of interest(PS-RoI)pooling with position-sensitive precise region of interest(PS-Pr-RoI)pooling,which avoids coordinate quantization and directly calculates two-order integrals for position-sensitive score maps,thus preventing a loss of spatial precision.A validation experiment was conducted in which the Microsoft common objects in context(MS COCO)training dataset was oversampled.Results showed an accuracy improvement of 3.7%for object detection tasks and an increase of 6.0%for small objects.
基金supported by the Key-Area Research and Development Program of Guangdong Province (2019B020214005)Agricultural Research Project and Agricultural Technology Promotion Project of Guangdong (2021KJ383)。
文摘Spike development directly affects the yield and quality of rice. We describe an algorithm for automatically identifying multiple developmental stages of rice spikes(AI-MDSRS) that transforms the automatic identification of multiple developmental stages of rice spikes into the detection of rice spikes of diverse maturity levels. The scales vary greatly in different growth and development stages because rice spikes are dense and small, posing challenges for their effective and accurate detection. We describe a rice spike detection model based on an improved faster regions with convolutional neural network(Faster R-CNN).The model incorporates the following optimization strategies: first, Inception_Res Net-v2 replaces VGG16 as a feature extraction network;second, a feature pyramid network(FPN) replaces single-scale feature maps to fuse with region proposal network(RPN);third, region of interest(Ro I) alignment replaces Ro I pooling, and distance-intersection over union(DIo U) is used as a standard for non-maximum suppression(NMS). The performance of the proposed model was compared with that of the original Faster R-CNN and YOLOv4 models. The mean average precision(m AP) of the rice spike detection model was92.47%, a substantial improvement on the original Faster R-CNN model(with 40.96% m AP) and 3.4%higher than that of the YOLOv4 model, experimentally indicating that the model is more accurate and reliable. The identification results of the model for the heading–flowering, milky maturity, and full maturity stages were within two days of the results of manual observation, fully meeting the needs of agricultural activities.
基金Departments of Medical Oncology and Radiation Oncology for study collaborationDepartment of Medical Record Library for medical record provisionthank Chinese Society of Clinical Oncology(CSCO) for partial financial support
文摘Background： The role of rituximab in combination with CHOP regimen in patients with stage I diffuse large B-cell lymphoma （DLBCL） remains to be defined. We aimed to compare CHOP plus rituximab （R-CHOP） with CHOP alone and determine the value of radiotherapy in these patients. Methods： Between 2003 and 2009, 140 untreated patients with stage I DLBCL were retrospectively analyzed in this study. Results： Seventy-eight patients were treated in R-CHOP group and 62 in CHOP group. Ninety-one patients received additional radiotherapy at the end of chemotherapy. The different treatment groups were well-balanced with respect to baseline characteristics. Complete response （CR） rate was 77% both in R-CHOP and CHOP groups （P=0.945）. After a median follow-up period of 56 months, patients received R-CHOP regimen had similar 5-year progression-free survival （PFS） （76% vs. 85%; log-rank P=0.215） and 5-year overall survival （OS） （90% vs. 96%; log-rank P=0.175） compared with those with CHOP alone. Patients with radiotherapy had significantly increased 5-year PFS compared with those who had chemotherapy alone （86% vs. 71%; log-rank P=0.005）. At multivariate analysis, patients who had CR （P=0.008） and received radiotherapy （P=0.003） were significantly associated with superior PFS. Conclusions： CHOP alone could be as effective as R-CHOP regimen and additional radiotherapy would be necessary for stage I or stage I non-bulky DLBCL patients.
文摘After defining the strong tensor product of strong (sub)chain complenes, it is shown that an analogue of the Kunneth theorem holds in strong homology by proving that the kernel (cokernel) of connecting homomorphisms is isomorphic to the direct sum of torsion (tensor) products of strong homology groups. An isomorphism between strong (r-stage) homology groups of inverse systems is also constructed.
文摘Data envelopment analysis(DEA)is a technique to measure the performance of decision-making units(DMUs).Conventional DEA treats DMUs as black boxes and the internal structure of DMUs is ignored.Two-stage DEA models are special case network DEA models that explore the internal structures of DMUs.Most often,one output cannot be produced by certain input data and/or the data may be expressed as ratio output/input.In these cases,traditional two-stage DEA models can no longer be used.To deal with these situations,we applied DEA-Ratio(DEA-R)to evaluate two-stage DMUs instead of traditional DEA.To this end,we developed two novel DEA-R models,namely,range directional DEA-R(RDD-R)and(weighted)Tchebycheff norm DEA-R(TND-R).The validity and reliability of our proposed approaches are shown by some examples.The Taiwan Residents non-life insurance companies are revisited using these proposed approaches and the results from the proposed methods are compared with those from some other methods.
