The ability to identify and temporally segment finegrained human actions throughout a video is crucial for robotics, surveillance, education, and beyond. Typical approaches decouple this problem by first extracting local spatiotemporal features from video frames and then feeding them into a temporal classifier that captures high-level temporal patterns. We introduce a new class of temporal models, which we call Temporal Convolutional Networks (TCNs), that use a hierarchy of temporal convolutions to perform fine-grained action segmentation or detection. Our Encoder-Decoder TCN uses pooling and upsampling to efficiently capture long-range temporal patterns whereas our Dilated TCN uses dilated convolutions. We show that TCNs are capable of capturing action compositions, segment durations, and long-range dependencies, and are over a magnitude faster to train than competing LSTM-based Recurrent Neural Networks. We apply these models to three challenging fine-grained datasets and show large improvements over the state of the art.
Recent progress in separating the speech signals from multiple overlapping speakers using a single audio channel has brought us closer to solving the cocktail party problem. However, most studies in this area use a constrained problem setup, comparing performance when speakers overlap almost completely, at artificially low sampling rates, and with no external background noise. In this paper, we strive to move the field towards more realistic and challenging scenarios. To that end, we created the WSJ0 Hipster Ambient Mixtures (WHAM!) dataset, consisting of two speaker mixtures from the wsj0-2mix dataset combined with real ambient noise samples. The samples were collected in coffee shops, restaurants, and bars in the San Francisco Bay Area, and are made publicly available. We benchmark various speech separation architectures and objective functions to evaluate their robustness to noise. While separation performance decreases as a result of noise, we still observe substantial gains relative to the noisy signals for most approaches.
Neuropathy, mechanical stress, and macrovascular disease are involved in the pathogenesis of diabetic foot ulceration. Implicit in the development of gangrene and ulceration is the recognition that these factors interact with the microcirculation, resulting in the failure of skin capillary flow to meet nutritive requirements. There is little evidence to associate structural microangiopathy with foot microcirculatory failure. Significant functional abnormalities of the microcirculation have been defined. In accord with the haemodynamic hypothesis early hyperaemia and capillary hypertension promote more sinister late functional abnormalities with increasing duration of diabetes. These late functional abnormalities include loss of autoregulation and reduced hyperaemic responses which interact with loss of neurogenic flow regulation, disturbed endothelial function, and abnormal rheology to produce the familiar clinical picture of the diabetic foot. Ischaemia secondary to multi-segment arterial disease induces additional abnormalities of microcirculatory function which are superimposed on the pre-existing diabetic microvascular structural and functional microangiopathy.
There is a close relationship between the abnormal microcirculation in diabetic subjects and diabetic neuropathy. Neurogenic factors play a prominent role in the regulation of the microcirculation. In diabetic neuropathy, damage to these mechanisms results in a profound haemodynamic disturbance with increased arteriovenous shunting, abnormal postural regulation of blood flow, and abnormal inflammatory responses to tissue injury. Abnormal neurogenic regulation of microvascular haemodynamics may contribute to the development of microangiopathy manifest as increased basement thickening and both are undoubtedly implicated in the pathogenesis of diabetic foot ulceration. In turn it is now recognized that microvascular abnormalities may contribute to the ischaemic aetiology of diabetic neuropathy.
The two major components of the microcirculation in the diabetic neuropathic foot have been examined in detail. Nutritive capillary blood flow was measured directly using the non-invasive technique of television microscopy, applied to the toe nailfold. Arteriovenous shunt flow was assessed using the technique of laser Doppler flowmetry, applied to the toe pulp. Fourteen diabetic patients with peripheral and autonomic neuropathy, 11 with no clinical evidence of neuropathy and 14 normal subjects were studied. Laser Doppler flowmetry (predominantly arteriovenous shunt flow) was increased more than three-fold (p less than 0.01) in the diabetic patients with neuropathy compared to control subjects, (median 3.57, interquartile range 2.00-5.32 volts vs median 0.93, interquartile range 0.47-2.36 volts, respectively). There was no evidence of skin capillary closure. The calculated capillary blood flow (erythrocyte flux) was significantly increased in the diabetic neuropathic patients compared to control subjects (median 76.4, interquartile range 34.4-109.8 picolitres/s vs median 23.2, range 8.0-44.8 picolitres/s, p less than 0.01). This study demonstrates that foot skin capillary blood flow is increased in diabetic patients with neuropathy. There is, therefore, no evidence to support the supposition that capillary ischaemia, either secondary to a "capillary steal phenomenon" or "advanced microangiopathy", is a feature of diabetic neuropathy under resting conditions.
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.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.