We propose a new technique for visual attribute transfer across images that may have very different appearance but have perceptually similar semantic structure. By visual attribute transfer, we mean transfer of visual information (such as color, tone, texture, and style) from one image to another. For example, one image could be that of a painting or a sketch while the other is a photo of a real scene, and both depict the same type of scene. Our technique finds semantically-meaningful dense correspondences between two input images. To accomplish this, it adapts the notion of "image analogy" [Hertzmann et al. 2001] with features extracted from a Deep Convolutional Neutral Network for matching; we call our technique deep image analogy. A coarse-to-fine strategy is used to compute the nearest-neighbor field for generating the results. We validate the effectiveness of our proposed method in a variety of cases, including style/texture transfer, color/style swap, sketch/painting to photo, and time lapse.

No abstract

No abstract

Figure 1: Our technique allows us to establish semantically-meaningful dense correspondences between two input images A and B . A and B are the reconstructed results subsequent to transfer of visual attributes.

On the basis of congener-specific analysis of dioxins in a dated sediment core, the sources and behavior of dioxins in Lake Shinji Basin, Japan, were estimated. The dioxins in the core showed that their deposition in the lake increased rapidly during the 1960s, peaked in the early 1970s, and then decreased gradually. Principal component analysis of the congener-specific data showed that three major sources existed: pentachlorophenol (PCP), chloronitrophen (CNP), and combustion. PCP and CNP are paddy field herbicides used extensively in the basin. The time trends of source contributions were estimated by multiple regression analysis using the source profiles. The results revealed that dioxin emission from PCP and CNP herbicides was high in the 1960s and the early 1970s, respectively. The contributions from PCP, CNP, and combustion in recent surface sediment were about 68, 16, and 16% in terms of total amount of dioxins. From the decreasing trend of dioxin deposition in the lake after extensive herbicide use, the amount of dioxins that accumulated in the agricultural soil in the basin was estimated to have decreased by about 2%/yr or a half-life of about 35 yr, indicating that dioxin runoff from agricultural fields would continue for a long time.

The Canadian Atmospheric Network for Currently Used Pesticides (CANCUP) was the first comprehensive, nationwide air surveillance study of pesticides in Canada. This paper presentsthe atmospheric occurrence and distribution of pesticides including organochlorine pesticides (OCPs), organophosphate pesticides (OPPs), acid herbicides (AHs), and neutral herbicides (NHs) during the spring to summer of 2004 and 2005 across agricultural regions in Canada. Atmospheric concentrations of pesticides varied within years and time periods, and regional characteristics were observed including the following: (i) highest air concentrations of several herbicides (e.g., mecoprop, triallate, and ethalfluralin) were found at Bratt's Lake, SK, a site in the Canadian Prairies; (ii) the west-coast site at Abbotsford, BC, had the maximum concentrations of diazinon; (iii) the fruit and vegetable growing region in Vineland, ON, showed highest levels for several insecticides including chlorpyrifos, endosulfan, and azinphos-methyl; (iv) high concentrations of atrazine and metolachlor were measured at St. Anicet, QC, a corn-growing region; (v) the Kensington site in PEI, Canada's largest potato-producing province, exhibited highest level of dimethoate. Analysis of particle- and gas-phase fractions of air samples revealed that most pesticides including OCPs, OPPs, and NHs exist mainly in the gas phase, while AHs exhibit more diversity in particle-gas partitioning behavior. This study also demonstrated that stirred up soil dust does not account for pesticides that are detected in the particle phase. The estimated dry and wet deposition fluxes indicate considerable atmospheric inputs for some current-use pesticides (CUPs). This data set represents the first measurements for many pesticides in the atmosphere, precipitation, and soil for given agricultural regions across Canada.

Heterogeneous domain adaptation (HDA) aims to facilitate the learning task in a target domain by borrowing knowledge from a heterogeneous source domain. In this paper, we propose a Soft Transfer Network (STN), which jointly learns a domain-shared classifier and a domain-invariant subspace in an end-to-end manner, for addressing the HDA problem. The proposed STN not only aligns the discriminative directions of domains but also matches both the marginal and conditional distributions across domains. To circumvent negative transfer, STN aligns the conditional distributions by using the soft-label strategy of unlabeled target data, which prevents the hard assignment of each unlabeled target data to only one category that may be incorrect. Further, STN introduces an adaptive coefficient to gradually increase the importance of the soft-labels since they will become more and more accurate as the number of iterations increases. We perform experiments on the transfer tasks of image-to-image, text-to-image, and text-to-text. Experimental results testify that the STN significantly outperforms several state-of-the-art approaches.

No abstract