دانلود رایگان مقاله انگلیسی مجموعه باز تشخیص فعالیت انسانی مبتنی بر امضا میکرو داپلر - الزویر 2019

Abstract

Open-set activity recognition remains as a challenging problem because of complex activity diversity. In previous works, extensive efforts have been paid to construct a negative set or set an optimal threshold for the target set. In this paper, a model based on Generative Adversarial Network (GAN), called ‘OpenGAN’ is proposed to address the open-set recognition without manual intervention during the training process. The generator produces fake target samples, which serve as an automatic negative set, and the discriminator is redesigned to output multiple categories together with an ‘unknown’ class. We evaluate the effectiveness of the proposed method on measured micro-Doppler radar dataset and the MOtion CAPture (MOCAP) database from Carnegie Mellon University (CMU). The comparison results with several state-of-the-art methods indicate that OpenGAN provides a promising open-set solution to human activity recognition even under the circumstance with few known classes. Ablation studies are also performed, and it is shown that the proposed architecture outperforms other variants and is robust on both datasets.

7. Conclusion and future work

In this paper, we design a novel model called ‘OpenGAN’ for open-set hu440 man activity detection and recognition. The model is inspired by generative adversarial networks and automatically constructs the negative set with synthesized samples from the generator without manual collection. Subsequently, the discriminator is modified with some reasonable modulations to adapt the open-set tasks. Both measurement radar dataset and MOCAP dataset are em445 ployed to verify the effectiveness of our model. Extensive experiments show that OpenGAN outperforms several comparison algorithms on the open-set human activity problem. The results suggest the robustness of our method against existing approaches. Exploration experiments about the network structure are also carried out. The results demonstrate that network with one sub-layer in each 450 dense block achieves the best performances under most circumstances. Future works will include extending the algorithm to other modal data and developing the model to an end-to-end open-set detection and recognition system.