We propose a deep learning approach for user-guided image colorization. The system directly maps a grayscale image, along with sparse, local user "hints" to an output colorization with a Convolutional Neural Network (CNN). Rather than using hand-defined rules, the network propagates user edits by fusing low-level cues along with high-level semantic information, learned from large-scale data. We train on a million images, with simulated user inputs. To guide the user towards efficient input selection, the system recommends likely colors based on the input image and current user inputs. The colorization is performed in a single feed-forward pass, enabling real-time use. Even with randomly simulated user inputs, we show that the proposed system helps novice users quickly create realistic colorizations, and offers large improvements in colorization quality with just a minute of use. In addition, we demonstrate that the framework can incorporate other user "hints" to the desired colorization, showing an application to color histogram transfer. Our code and models are available at https://richzhang.github.io/ideepcolor.
We propose split-brain autoencoders, a straightforward modification of the traditional autoencoder architecture, for unsupervised representation learning. The method adds a split to the network, resulting in two disjoint sub-networks. Each sub-network is trained to perform a difficult taskpredicting one subset of the data channels from another. Together, the sub-networks extract features from the entire input signal. By forcing the network to solve crosschannel prediction tasks, we induce a representation within the network which transfers well to other, unseen tasks. This method achieves state-of-the-art performance on several large-scale transfer learning benchmarks.
Proprotein convertase subtilisin/kexin type 9 (PCSK9) regulates serum LDL cholesterol (LDL-C) by interacting with the LDL receptor (LDLR) and is an attractive therapeutic target for LDL-C lowering. We have generated a neutralizing anti-PCSK9 antibody, mAb1, that binds to an epitope on PCSK9 adjacent to the region required for LDLR interaction. In vitro, mAb1 inhibits PCSK9 binding to the LDLR and attenuates PCSK9-mediated reduction in LDLR protein levels, thereby increasing LDL uptake. A combination of mAb1 with a statin increases LDLR levels in HepG2 cells more than either treatment alone. In wild-type mice, mAb1 increases hepatic LDLR protein levels Ϸ2-fold and lowers total serum cholesterol by up to 36%: this effect is not observed in LDLR ؊/؊ mice. In cynomolgus monkeys, a single injection of mAb1 reduces serum LDL-C by 80%, and a significant decrease is maintained for 10 days. We conclude that anti-PCSK9 antibodies may be effective therapeutics for treating hypercholesterolemia.antibody ͉ LDL-C ͉ LDLR ͉ PCSK9 ͉ hypercholesterolemia P roprotein convertase subtilisin/kexin type 9 (PCSK9) has been implicated as an important regulator of LDL metabolism (1, 2). Human genetic studies provide strong validation for the role of PCSK9 in modulating LDL cholesterol (LDL-C) levels and the incidence of coronary heart disease (CHD) in man. Gain-of-function (GOF) mutations in the PCSK9 gene are associated with elevated serum LDL-C levels (Ͼ300 mg/dL) and premature CHD (3), whereas loss-of-function (LOF) mutations are associated with low serum LDL-C (Յ100 mg/dL) (4). Strikingly, subjects harboring the heterozygous LOF mutations exhibited an 88% reduction in the incidence of CHD over a 15-year period relative to noncarriers of the mutations (5). Moreover, despite a complete loss of PCSK9 and serum LDL-C of Ͻ20 mg/dL, the 2 subjects carrying compound heterozygote LOF mutations appear healthy (6, 7).PCSK9 belongs to the subtilisin family of serine proteases and consists of a prodomain, catalytic domain, and C-terminal V domain (8). Expressed highly in the liver, PCSK9 is secreted after autocatalytic cleavage of its zymogen form (1). The prodomain remains noncovalently associated with the catalytic domain and seems to inhibit further proteolytic enzyme activity (8, 9). Secreted PCSK9 modulates LDL-C levels by posttranslational downregulation of hepatic LDL receptor (LDLR) protein (1). The precise mechanism is unknown, but a direct interaction between repeat A of the LDLR EGF homology domain and the PCSK9 catalytic domain is required (10, 11). Proteolytic cleavage of the LDLR by PCSK9 does not occur (12, 13); rather, the PCSK9:LDLR complex is endocytosed and directed to the endosome/lysosome compartment for degradation (14, 15). Current understanding of the LDLR pathway asserts that apolipoprotein B (apoB) and E (apoE) containing lipoprotein particles endocytosed with the LDLR are transported to the acidic environment of the endosome, where they dissociate from the receptor and are subsequently catabolized in lysosomes, while t...
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.