No abstract

No abstract

Two-person interaction recognition has become an area of growing interest in human action recognition. The graph convolutional network (GCN) using human skeleton data has been shown to be highly effective for action recognition. Most GCN-based methods focus on recognizing an individual person's actions on the basis of an intra-body graph. However, many of these methods do not represent the relation between two bodies, making it difficult to accurately recognize human interaction. In this work, we propose multi-stream adaptive GCN using inter-and intra-body graphs (MAGCN-IIG) as a new method of human interaction recognition. To ensure highly accurate human interaction recognition, our method cooperatively utilizes two types of graphs: an inter-body graph and an intra-body graph. The inter-body graph, which is newly introduced in this paper, connects the inter-body joints between two people as well as intra-body connections. The adaptive GCN using the inter-body graph captures the relation of joints between two people, even different types of joints located far away from each other. Further, by implementing a multi-stream architecture, our method simultaneously captures both inter-body and intra-body relations in each of two units that represent the position and motion of people. Experiments on interaction recognition using two large-scale human action datasets, NTU RGB+D and NTU RGB+D 120, showed that our method recognized human interactions more accurately than state-of-the-art methods.INDEX TERMS Graph convolutional network, human interaction recognition, multi-stream network, skeleton-based action recognition.

ABSTRACT:Semi-rigid polyesters based on twin biphenyl analogues of 1,3,4-thiadiazole (2-phenyl-1,3,4-thiadiazoles) with hepta-and octamethylene chains in the central part were synthesized by melt polycondensation of bismethyl ester derivatives of twin 2-phenyl-1,3,4-thiadiazole with a ,w-alkylenediols and dioxydialcohol derivatives of bi phenyl. The relationship between polymer structure and thermotropic liquid-crystallinity (LC) was investigated by differential scanning calorimetry (DSC), polarizing microscope observations, miscibility tests and temperature-dependent X-ray analysis. All the polymers derived from a ,w-alkylenediols appeared to form enantiotropic or monotropic nematic phase independent of the lengths of aliphatic chains. Among the polyesters containing the bi phenyl unit, most polymers with octamethylene segment between the twin 2-phenyl-1,3,4-thiadiazoles showed the nematic phase, but polymers composed of the twin bi phenyl analogue having the central heptamethylene chain had no LC melts.KEY WORDS Semi-Rigid Polyester/ 2-Phenyl-1,3,4-thiadiazole / Melt Polycondensation / Bismethyl Ester Derivative/ Liquid-Crystallinity / Nematic Phase/ Differential Scanning Calorimetry Measurement/ A terphenyl analogue of 1,3,4-thiadiazole (2,5-diphenyl-1,3,4-thiadiazole), a five-membered heterocyclic unit containing sulfur and nitrogen atoms, is a calamitic mesogen. 1 Derivatives composed of mesogen with flexible chains on both sides, twin terphenyl analogues of the 1,3,4-thiadiazole having alkylenedioxy segments in the center 2 -5 and main-6 -8 or side-chain 9 polymers containing the mesogen form nematic and smectic mesophases in spite of its bent molecular structure. 2 · 10 -12 Our previous articles described that twin 2,5-diphenyl-1,3,4-thiadiazole derivatives linking an octamethylenedioxy chain to the center 5 have stable nematic and smectic phases. Side-chain polymers with the 2,5-diphenyl-1,3,4-thiadiazole as a mesogen represent cholesteric and chiral smectic C * phases.9 Enantiotropic thermotropic liquid-crystalline (LC) phases emerge in semi-rigid polyesters derived from diacylchloride of the 2,5-diphenyl-1,3,4-thiadiazole and aliphatic diols, 8 although wholly aromatic polyesters consisting of calamitic mesogen have no LC melts. 13Biphenyl analogues of 1,3,4-thiadiazole (2-phenyl-1,3,4-thiadiazoles) have the monotropic smectic A phase, 1 · 14 · 15 although LC temperature range was narrow. 2 In recent years, Tschierske et al. found that homoand heterotrimeric or tetrameric derivatives with the 2-phenyl-1,3,4-thiadiazole ring linked to central flexible chains show the smectic C phase. 14 · 15 Thermotropic LC phases appear in semi-rigid LC polyesters based on twin biphenyl analogue of 1,3,4-thiadiazole having decamethylene chain in the central part. 16 The polymer with octamethylene spacer displayed the monotropic nematic phase and those having decamethylene and dodecamethylene chains showed enantiotropic smectic phases, but LC temperature range was narrow.The present work describes the synthesis of vario...

We present Cycle-Contrastive Learning (CCL), a novel self-supervised method for learning video representation. Following a nature that there is a belong and inclusion relation of video and its frames, CCL is designed to find correspondences across frames and videos considering the contrastive representation in their domains respectively. It is different from recent approaches that merely learn correspondences across frames or clips. In our method, the frame and video representations are learned from a single network based on an R3D architecture, with a shared non-linear transformation for embedding both frame and video features before the cycle-contrastive loss. We demonstrate that the video representation learned by CCL can be transferred well to downstream tasks of video understanding, outperforming previous methods in nearest neighbour retrieval and action recognition tasks on UCF101, HMDB51 and MMAct.Preprint. Under review.