Siamese networks have drawn great attention in visual tracking because of their balanced accuracy and speed. However, the backbone networks used in Siamese trackers are relatively shallow, such as AlexNet [18], which does not fully take advantage of the capability of modern deep neural networks. In this paper, we investigate how to leverage deeper and wider convolutional neural networks to enhance tracking robustness and accuracy. We observe that direct replacement of backbones with existing powerful architectures, such as ResNet [14] and Inception [33], does not bring improvements. The main reasons are that 1) large increases in the receptive field of neurons lead to reduced feature discriminability and localization precision; and 2) the network padding for convolutions induces a positional bias in learning. To address these issues, we propose new residual modules to eliminate the negative impact of padding, and further design new architectures using these modules with controlled receptive field size and network stride. The designed architectures are lightweight and guarantee real-time tracking speed when applied to SiamFC [2] and SiamRPN [20]. Experiments show that solely due to the proposed network architectures, our SiamFC+ and SiamRPN+ obtain up to 9.8%/5.7% (AUC), 23.3%/8.8% (EAO) and 24.4%/25.0% (EAO) relative improvements over the original versions [2, 20] on the OTB-15, VOT-16 and VOT-17 datasets, respectively.

No abstract

QUASAR arterial spin labeling (ASL) was used to investigate the role of vascular impairment in Alzheimer's disease (AD). We hypothesized that the hemodynamic parameters monitoring cerebrovascular integrity, i.e., cerebral blood flow (CBF), arterial blood volume (aBV), and arterial transit time (aTT), would be affected. 13 AD patients and 15 healthy control (HC) subjects underwent 3T MRI scanning. Two separate blood flow acquisitions were obtained with 1 slice overlap for whole brain coverage. CBF, aBV, and aTT maps were calculated using in-house software. Preprocessing and statistical analyses were performed on SPM5. Region-of-interest (ROI) studies of ten selected cerebral regions were also conducted. There were significant differences in mini mental status exam (MMSE) (AD: 16.3 ± 4.55, HC: 28.5 ± 2.00) and Alzheimer's disease assessment scale-cognitive subscale (ADAS-cog) scores (AD: 25.25 ± 9.64, HC: 5.51 ± 2.62) between the 2 groups (p < 0.001) but none in age (p = 0.068). CBF decreased significantly (p < 0.01) in AD compared to controls in the right middle cingulate, left cuneus, left inferior and middle frontal, right superior frontal, left inferior parietal, and right supramarginal gyri. ROI studies confirmed significant hemodynamic impairments in AD compared to HC (p < 0.05): CBF in middle and posterior cingulate, aBV in left superior temporal, right inferior parietal, and posterior cingulate, and aTT in left inferior frontal and middle cingulate gyri. CBF correlated positively while aTT correlated negatively to MMSE, and vice versa for ADAS-cog. Using QUASAR ASL, we found patterns of regional hemodynamic impairment typical of moderate AD, suggesting underlying vascular abnormality. As potential biomarkers, these hemodynamic parameters could differentiate patients from volunteers, and possibly indicate the conversion from healthy aging to mild cognitive impairment to AD.

Purpose The purpose of this paper is to investigate whether firms with high environmental disclosure have a low possibility of non-standard audit opinions and audit fees and whether this trend is more obvious after than prior to the Measures for the Disclosure of Environmental Information (Measure) implemented in 2008. Design/methodology/approach Based on the Measures implemented in 2008, the authors select data for the listed manufacturing firms from 2004 to 2006 (Pre-Measure) and from 2009 to 2011 (Post-Measure) as research samples to investigate the relationships between environmental disclosures, audit opinions and audit fees with difference in difference models. In addition, we also consider the influence of media attention, the polluting industry and internal control on the audit effect of environmental disclosure. Findings The results show that the level of environmental disclosure is significantly negatively correlated with the possibility of issuing non-standard audit opinions and audit fees after measure is implemented, especially hard environmental information. Further evidence indicates that the auditing effect of environmental disclosures is stronger on firms that receive less media attention, in firms with better internal controls, and in firms belonging to industries with heavy pollution. Originality/value In the Chinese setting, a high level of environmental information disclosures can effectively reduce the audit risk and lead to a high possibility of standard audit opinions and low audit fees. This effect is pronounced after issuing Measure. The conclusions suggest that measure and increasing environmental disclosure have an obvious positive audit effect and that firms should be forced or encouraged to disclose more environmental information from the perspective of auditors in China.

A coal burst mitigation strategy for tailgate in mining of deep inclined longwall panels with top coal caving at Huafeng Coal Mine is presented in this paper. Field data showed that coal bursts, rib sloughing or slabbing, large convergence, and so forth frequently occurred within the tailgate entries during development and panel retreating employing standard longwall top coal caving (LTCC) layout which resulted in fatal injuries and tremendous profit loss. The contributing factors leading to coal bursts were analyzed. Laboratory tests, in situ measurement, and field observation demonstrate that the intrinsic bursting proneness of the coal seam and immediate roof stratum, deep cover, overlying ultrathick (500–800 m) conglomerate strata, faults, and, most importantly, improper panel layout led to coal bursts. By employing a new strategy, that is, longwall mining with split-level gateroads (LMSG), gateroads on either end of a LMSG panel are located at different levels within a coal seam, adjacent LMSG panels overlap end to end, and the tailgate of the adjacent new LMSG panel can be located below the headgate entry of the previous LMSG panel or may be offset horizontally with respect to it. Numerical modeling was carried out to investigate the stress distribution and yield zone development within surrounding rock mass which was validated by field investigation. The results indicate that standard LTCC system gave rise to high ground pressure around tailgate entries next to the gob, while LMSG tailgate entry below the gob edge was in a destressed environment. Therefore, coal bursts are significantly mitigated. Field practice of LMSG at Huafeng Coal Mine demonstrates how the new strategy effectively dealt with coal burst problems in mining of deep inclined longwall panels with a reduced incidence of ground control problems. The new strategy can potentially be applied in similar settings.