Makeup transfer: A review
Makeup transfer (MT) aims to transfer the makeup style from a given reference makeup face image to a source image while preserving face identity and background information. In recent years, MT has attracted the attention of many scholars, and it ...
Denseformer: A dense transformer framework for person re‐identification
Transformer has shown its effectiveness and advantage in many computer vision tasks, for example, image classification and object re‐identification (ReID). However, existing vision transformers are stacked layer by layer, lacking direct ...
ACGAN: Age‐compensated makeup transfer based on homologous continuity generative adversarial network model
The authors focus on the makeup transformation problem, which refers to the transfer of makeup from a reference face to a source face image while maintaining the source makeup‐free face image. In recent years, makeup transformation has become a ...
Global centralised and structured discriminative non‐negative matrix factorisation for hyperspectral unmixing
Advances have been achieved in hyperspectral unmixing using the existing manifold Non‐negative Matrix Factorisation methods, although most of these methods only exploit the preliminary structural information, that is, the nearest neighbour graph. ...
The proposed GSNMF holds the power to guarantee those similar raw data having similar abundance, and simultaneously ensure dissimilar data has different estimated abundance. In addition, GSNMF can effectively acquire the global structure characteristics. ...
A dual‐balanced network for long‐tail distribution object detection
Object detection on datasets with imbalanced distributions (i.e. long‐tail distributions) dataset is a significantly challenging task. Some re‐balancing solutions, such as re‐weighting and re‐sampling have two main disadvantages. First, re‐...
In this paper, a dual‐balanced network is proposed for long‐tail distribution object detection. The proposal in RPN and the logic in classification networks are balanced, which reduces the AP difference between the head and tail categories and improves ...
Point completion by a Stack‐Style Folding Network with multi‐scaled graphical features
Point cloud completion is prevalent due to the insufficient results from current point cloud acquisition equipments, where a large number of point data failed to represent a relatively complete shape. Existing point cloud completion algorithms, ...
We propose a novel end‐to‐end architecture for point completion using a stack‐style folding network called the SSFN. Experiments on ShapeNet and KITTI indicate that our SSFN achieves a decent vision quality and metric performance. image image
Few‐shot logo detection
The proliferation of deep learning has driven research into deep learning‐based logo detection, which usually needs a large number of annotated data to train the model. However, due to the occasional appearance of new brands or the high cost of ...
The authors balanced feature pyramid and deformable RoI pooling in DCH‐FSLogo, this enhances the feature extraction capability and adapts to the different logo shapes. In addition, the authors introduce the double classification head for few‐shot logo ...
Zero‐shot temporal event localisation: Label‐free, training‐free, domain‐free
Temporal event localisation (TEL) has recently attracted increasing attention due to the rapid development of video platforms. Existing methods are based on either fully/weakly supervised or unsupervised learning, and thus they rely on expensive ...
The authors propose a ZSTEL method which can work without training data or annotations. Leveraging large‐scale vision and language pre‐trained models, for example, CLIP, the two key problems are solved: (1) how to find the relevant region where the event ...