A pre-trained 2D CNN (Convolutional Neural Network) can be used for the spatial stream in the two-stream CNN structure for videos, treating the representative frame selected from the video as an input. However, the CNN for the temporal stream in the two-stream CNN needs training from scratch using the optical flow frames, which demands expensive computations. In this paper, we propose to adopt a pre-trained 2D CNN for the temporal stream to avoid the optical flow computations. Specifically, three RGB frames selected at three different times in the video sequence are converted into grayscale images and are assigned to three R(red), G(green), and B(blue) channels, respectively, to form a Stacked Grayscale 3channel Image (SG3I). Then, the pre-trained 2D CNN is fine-tuned by SG3Is for the temporal stream CNN. Therefore, only pre-trained 2D CNNs are used for both spatial and temporal streams. To learn long-range temporal motions in videos, we can use multiple SG3Is by partitioning the video shot into sub-shots and a single SG3I is generated for each sub-shot. Experimental results show that our two-stream CNN with the proposed SG3Is is about 14.6 times faster than the first version of the two-stream CNN with the optical flow, and yet achieves a similar recognition accuracy for UCF-101 and a 5.7% better result for HMDB-51. INDEX TERMS Convolutional neural network (CNN), action recognition, video analysis, two-stream convolutional neural networks.

SMD (Singapore Maritime Dataset) is a public dataset with annotated videos, and it is almost unique in the training of deep neural networks (DNN) for the recognition of maritime objects. However, there are noisy labels and imprecisely located bounding boxes in the ground truth of the SMD. In this paper, for the benchmark of DNN algorithms, we correct the annotations of the SMD dataset and present an improved version, which we coined SMD-Plus. We also propose augmentation techniques designed especially for the SMD-Plus. More specifically, an online transformation of training images via Copy & Paste is applied to solve the class-imbalance problem in the training dataset. Furthermore, the mix-up technique is adopted in addition to the basic augmentation techniques for YOLO-V5. Experimental results show that the detection and classification performance of the modified YOLO-V5 with the SMD-Plus has improved in comparison to the original YOLO-V5. The ground truth of the SMD-Plus and our experimental results are available for download.

BackgroundThis study was performed to evaluate their 5-year survival rates and identify the factors affecting the prognosis of oral cancer patients who had undergone surgical treatment only.MethodsAmong 130 patients who were diagnosed with malignant tumor of oral, maxillofacial, and surgical treated in the Department of Oral and Maxillofacial Surgery at Chonnam National University Hospital within a period from January 2000 to December 2010, for 11 years, 84 patients were investigated who were followed up for more than 5 years after radical surgery; oral cancer is primary and received only surgical treatment. The survival rate according to gender, age, type and site of cancer, TNM stage, cervical lymph node metastasis and its stage, recurrence or metastasis, time of recurrence and metastasis, and differentiation were investigated and analyzed.ResultsOverall, 5-year survival rate in patients who received only surgical treatment was 81.2 %, and disease-specific 5-year survival rate was 83.1 %. The disease-specific 5-year survival rate based on TNM stage, metastasis of cervical lymph node, N stage, and presence of recurrence/metastasis was a significant difference (p < 0.05). The disease-specific 5-year survival rate based on sex, age, type of tumor, primary site, and differentiation was not a significant difference (p > 0.05).ConclusionsThese results suggest that good survival rate can be obtained with surgical treatment only, and stage of oral cancer, cervical lymph node metastasis and stage, recurrence or metastasis, time of recurrence, and metastasis have a significant effect on survival rate in oral cancer patients.

No abstract

BackgroundThis study was performed to evaluate three-dimensional positional change of the condyle using three-dimensional computed tomography (3D-CT) following unilateral sagittal split ramus osteotomy (USSRO) in patients with mandibular prognathism.MethodsThis study examined two patients exhibiting skeletal class III malocclusion with facial asymmetry who underwent USSRO for a mandibular setback. 3D-CT was performed before surgery, immediately after surgery, and 6 months postoperatively.After creating 3D-CT images by using the In-vivo 5™ program, the axial plane, coronal plane, and sagittal plane were configured. Three-dimensional positional changes from each plane to the condyle, axial condylar head axis angle (AHA), axial condylar head position (AHP), frontal condylar head axis angle (FHA), frontal condylar head position (FHP), sagittal condylar head axis angle (SHA), and sagittal condylar head position (SHP) of the two patients were measured before surgery, immediately after surgery, and 6 months postoperatively.ResultsIn the first patient, medial rotation of the operated condyle in AHA and anterior rotation in SHA were observed. There were no significant changes after surgery in AHP, FHP, and SHP after surgery. In the second patient, medial rotation of the operated condyle in AHA and lateral rotation of the operated condyle in FHA were observed. There were no significant changes in AHP, FHP, and SHP postoperatively. This indicates that in USSRO, postoperative movement of the condylar head is insignificant; however, medial rotation of the condylar head is possible. Although three-dimensional changes were observed, these were not clinically significant.ConclusionsThe results of this study suggest that although three-dimensional changes in condylar head position are observed in patients post SSRO, there are no significant changes that would clinically affect the patient.