In this report, we present a new face detection scheme using deep learning and achieve the state-of-the-art detection performance on the well-known FDDB face detetion benchmark evaluation. In particular, we improve the state-of-the-art faster RCNN framework by combining a number of strategies, including feature concatenation, hard negative mining, multi-scale training, model pretraining, and proper calibration of key parameters. As a consequence, the proposed scheme obtained the state-of-the-art face detection performance, making it the best model in terms of ROC curves among all the published methods on the FDDB benchmark.
Whether there is a general allometry law across plant species with different sizes and under different environment has long been controversial and shrubs are particularly useful to examine these questions. Here we sampled 939 individuals from 50 forest shrub species along a large altitudinal gradient. We tested several allometry models with four relationships simultaneously (between stem diameter, height, leaf, stem and aboveground biomass), including geometric, elastic and stress similarity, and metabolic scaling theory’s predictions on small plants (MSTs) and trees (MSTt). We also tested if allometric exponents change markedly with climate and phylogeny. The predicted exponents of MSTt, elastic similarity and stress similarity (models for trees) were not supported by our data, while MSTs and geometric similarity gained more support, suggesting the finite size effect is more important for shrub allometries than being a woody plant. The influence of climate and phylogeny on allometric exponents were not significant or very weak, again suggesting strong biophysical constraints on shrub allometries. Our results reveal clear differences of shrub allometries from previous findings on trees (e.g. much weaker climatic and phylogenic control). Comparisons of herbs, shrubs and trees along a same climatic gradient are needed for better understanding of plant allometries.
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.