In this paper, we present a detailed design of dynamic video segmentation network (DVSNet) for fast and efficient video semantic segmentation. DVSNet consists of two convolutional neural networks: a segmentation network and a flow network. The former generates highly accurate semantic segmentations, but is deeper and slower. The latter is much faster than the former, but its output requires further processing to generate less accurate semantic segmentations. We explore the use of a decision network to adaptively assign different frame regions to different networks based on a metric called expected confidence score. Frame regions with a higher expected confidence score traverse the flow network. Frame regions with a lower expected confidence score have to pass through the segmentation network. We have extensively performed experiments on various configurations of DVSNet, and investigated a number of variants for the proposed decision network. The experimental results show that our DVSNet is able to achieve up to 70.4% mIoU at 19.8 fps on the Cityscape dataset. A high speed version of DVSNet is able to deliver an fps of 30.4 with 63.2% mIoU on the same dataset. DVSNet is also able to reduce up to 95% of the computational workloads.
Osteosarcoma is characterized by a high malignant and metastatic potential. The chemokine stromal-derived factor-1alpha (SDF-1alpha) and its receptor, CXCR4, play a crucial role in adhesion and migration of human cancer cells. Integrins are the major adhesive molecules in mammalian cells, and has been associated with metastasis of cancer cells. Here, we found that human osteosarcoma cell lines had significant expression of SDF-1 and CXCR4 (SDF-1 receptor). Treatment of osteosarcoma cells with SDF-1alpha increased the migration and cell surface expression of alphavbeta3 integrin. CXCR4-neutralizing antibody, CXCR4 specific inhibitor (AMD3100) or small interfering RNA against CXCR4 inhibited the SDF-1alpha-induced increase the migration and integrin expression of osteosarcoma cells. Pretreated of osteosarcoma cells with MAPK kinase (MEK) inhibitor PD98059 inhibited the SDF-1alpha-mediated migration and integrin expression. Stimulation of cells with SDF-1alpha increased the phosphorylation of MEK and extracellular signal-regulating kinase (ERK). In addition, NF-kappaB inhibitor (PDTC) or IkappaB protease inhibitor (TPCK) also inhibited SDF-1alpha-mediated cell migration and integrin up-regulation. Stimulation of cells with SDF-1alpha induced IkappaB kinase (IKKalpha/beta) phosphorylation, IkappaB phosphorylation, p65 Ser(536) phosphorylation, and kappaB-luciferase activity. Furthermore, the SDF-1alpha-mediated increasing kappaB-luciferase activity was inhibited by AMD3100, PD98059, PDTC and TPCK or MEK1, ERK2, IKKalpha and IKKbeta mutants. Taken together, these results suggest that the SDF-1alpha acts through CXCR4 to activate MEK and ERK, which in turn activates IKKalpha/beta and NF-kappaB, resulting in the activations of alphavbeta3 integrins and contributing the migration of human osteosarcoma cells.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.