Detection and recognition of scene texts of arbitrary shapes remain a grand challenge due to the super-rich text shape variation in text line orientations, lengths, curvatures, etc. This paper presents a mask-guided multi-task network that detects and rectifies scene texts of arbitrary shapes reliably. Three types of keypoints are detected which specify the centre line and so the shape of text instances accurately. In addition, four types of keypoint links are detected of which the horizontal links associate the detected keypoints of each text instance and the vertical links predict a pair of landmark points (for each keypoint) along the upper and lower text boundary, respectively. Scene texts can be located and rectified by linking up the associated landmark points (giving localization polygon boxes) and transforming the polygon boxes via thin plate spline, respectively. Extensive experiments over several public datasets show that the use of text keypoints is tolerant to the variation in text orientations, lengths, and curvatures, and it achieves superior scene text detection and rectification performance as compared with state-of-the-art methods.
Scene understanding based on LiDAR point cloud is an essential task for autonomous cars to drive safely, which often employs spherical projection to map 3D point cloud into multi-channel 2D images for semantic segmentation. Most existing methods simply stack different point attributes/modalities (e.g. coordinates, intensity, depth, etc.) as image channels to increase information capacity, but ignore distinct characteristics of point attributes in different image channels. We design FPS-Net, a convolutional fusion network that exploits the uniqueness and discrepancy among the projected image channels for optimal point cloud segmentation. FPS-Net adopts an encoder-decoder structure. Instead of simply stacking multiple channel images as a single input, we group them into different modalities to first learn modality-specific features separately and then map the learned features into a common high-dimensional feature space for pixel-level fusion and learning. Specifically, we design a residual dense block with multiple receptive fields as a building block in the encoder which preserves detailed information in each modality and learns hierarchical modality-specific and fused features effectively. In the FPS-Net decoder, we use a recurrent convolution block likewise to hierarchically decode fused features into output space for pixel-level classification. Extensive experiments conducted on two widely adopted point cloud datasets show that FPS-Net achieves superior semantic segmentation as compared with state-of-the-art projection-based methods. In addition, the proposed modality fusion idea is compatible with typical projection-based methods and can be incorporated into them with consistent performance improvements.
Unsupervised domain adaptive object detection aims to adapt detectors from a labelled source domain to an unlabelled target domain. Most existing works take a two-stage strategy that first generates region proposals and then detects objects of interest, where adversarial learning is widely adopted to mitigate the inter-domain discrepancy in both stages. However, adversarial learning may impair the alignment of well-aligned samples as it merely aligns the global distributions across domains. To address this issue, we design an uncertainty-aware domain adaptation network (UaDAN) that introduces conditional adversarial learning to align well-aligned and poorly-aligned samples separately in different manners. Specifically, we design an uncertainty metric that assesses the alignment of each sample and adjusts the strength of adversarial learning for well-aligned and poorly-aligned samples adaptively. In addition, we exploit the uncertainty metric to achieve curriculum learning that first performs easier image-level alignment and then more difficult instance-level alignment progressively. Extensive experiments over four challenging domain adaptive object detection datasets show that UaDAN achieves superior performance as compared with state-of-the-art methods.
Lighting estimation from a single image is an essential yet challenging task in computer vision and computer graphics. Existing works estimate lighting by regressing representative illumination parameters or generating illumination maps directly. However, these methods often suffer from poor accuracy and generalization. This paper presents Geometric Mover's Light (GMLight), a lighting estimation framework that employs a regression network and a generative projector for effective illumination estimation. We parameterize illumination scenes in terms of the geometric light distribution, light intensity, ambient term, and auxiliary depth, and estimate them as a pure regression task. Inspired by the earth mover's distance, we design a novel geometric mover's loss to guide the accurate regression of light distribution parameters. With the estimated lighting parameters, the generative projector synthesizes panoramic illumination maps with realistic appearance and frequency. Extensive experiments show that GMLight achieves accurate illumination estimation and superior fidelity in relighting for 3D object insertion.
Illumination estimation from a single image is critical in 3D rendering and it has been investigated extensively in the computer vision and computer graphic research community. On the other hand, existing works estimate illumination by either regressing light parameters or generating illumination maps that are often hard to optimize or tend to produce inaccurate predictions. We propose Earth Mover Light (EMLight), an illumination estimation framework that leverages a regression network and a neural projector for accurate illumination estimation. We decompose the illumination map into spherical light distribution, light intensity and the ambient term, and define the illumination estimation as a parameter regression task for the three illumination components. Motivated by the Earth Mover distance, we design a novel spherical mover's loss that guides to regress light distribution parameters accurately by taking advantage of the subtleties of spherical distribution. Under the guidance of the predicted spherical distribution, light intensity and ambient term, the neural projector synthesizes panoramic illumination maps with realistic light frequency. Extensive experiments show that EMLight achieves accurate illumination estimation and the generated relighting in 3D object embedding exhibits superior plausibility and fidelity as compared with state-of-the-art methods.
Dealing with the inconsistency between a foreground object and a background image is a challenging task in high-fidelity image composition. State-of-the-art methods strive to harmonize the composed image by adapting the style of foreground objects to be compatible with the background image, whereas the potential shadow of foreground objects within the composed image which is critical to the composition realism is largely neglected. In this paper, we propose an Adversarial Image Composition Net (AIC-Net) that achieves realistic image composition by considering potential shadows that the foreground object projects in the composed image. A novel branched generation mechanism is proposed, which disentangles the generation of shadows and the transfer of foreground styles for optimal accomplishment of the two tasks simultaneously. A differentiable spatial transformation module is designed which bridges the local harmonization and the global harmonization to achieve their joint optimization effectively. Extensive experiments on pedestrian and car composition tasks show that the proposed AIC-Net achieves superior composition performance qualitatively and quantitatively.
Recent studies on facial expression editing have obtained very promising progress. On the other hand, existing methods face the constraint of requiring a large amount of expression labels which are often expensive and time-consuming to collect. This paper presents an innovative label-free expression editing via disentanglement (LEED) framework that is capable of editing the expression of both frontal and profile facial images without requiring any expression label. The idea is to disentangle the identity and expression of a facial image in the expression manifold, where the neutral face captures the identity attribute and the displacement between the neutral image and the expressive image captures the expression attribute. Two novel losses are designed for optimal expression disentanglement and consistent synthesis, including a mutual expression information loss that aims to extract pure expression-related features and a siamese loss that aims to enhance the expression similarity between the synthesized image and the reference image. Extensive experiments over two public facial expression datasets show that LEED achieves superior facial expression editing qualitatively and quantitatively.
Recent advances in unsupervised domain adaptation for semantic segmentation have shown great potentials to relieve the demand of expensive per-pixel annotations. However, most existing works address the domain discrepancy by aligning the data distributions of two domains at a global image level whereas the local consistencies are largely neglected. This paper presents an innovative local contextual-relation consistent domain adaptation (CrCDA) technique that aims to achieve local-level consistencies during the global-level alignment. The idea is to take a closer look at region-wise feature representations and align them for local-level consistencies. Specifically, CrCDA learns and enforces the prototypical local contextual-relations explicitly in the feature space of a labelled source domain while transferring them to an unlabelled target domain via backpropagation-based adversarial learning. An adaptive entropy max-min adversarial learning scheme is designed to optimally align these hundreds of local contextual-relations across domain without requiring discriminator or extra computation overhead. The proposed CrCDA has been evaluated extensively over two challenging domain adaptive segmentation tasks (e.g., GTA5 to Cityscapes and SYNTHIA to Cityscapes), and experiments demonstrate its superior segmentation performance as compared with state-of-the-art methods.
Recent advances in generative adversarial networks (GANs) have achieved great success in automated image composition that generates new images by embedding interested foreground objects into background images automatically. On the other hand, most existing works deal with foreground objects in two-dimensional (2D) images though foreground objects in three-dimensional (3D) models are more flexible with 360-degree view freedom. This paper presents an innovative View Alignment GAN (VA-GAN) that composes new images by embedding 3D models into 2D background images realistically and automatically. VA-GAN consists of a texture generator and a differential discriminator that are inter-connected and end-to-end trainable. The differential discriminator guides to learn geometric transformation from background images so that the composed 3D models can be aligned with the background images with realistic poses and views. The texture generator adopts a novel view encoding mechanism for generating accurate object textures for the 3D models under the estimated views. Extensive experiments over two synthesis tasks (car synthesis with KITTI and pedestrian synthesis with Cityscapes) show that VA-GAN achieves high-fidelity composition qualitatively and quantitatively as compared with state-of-the-art generation methods.
Often the best performing deep neural models are ensembles of multiple base-level networks, nevertheless, ensemble learning with respect to domain adaptive person re-ID remains unexplored. In this paper, we propose a multiple expert brainstorming network (MEB-Net) for domain adaptive person re-ID, opening up a promising direction about model ensemble problem under unsupervised conditions. MEB-Net adopts a mutual learning strategy, where multiple networks with different architectures are pre-trained within a source domain as expert models equipped with specific features and knowledge, while the adaptation is then accomplished through brainstorming (mutual learning) among expert models. MEB-Net accommodates the heterogeneity of experts learned with different architectures and enhances discrimination capability of the adapted re-ID model, by introducing a regularization scheme about authority of experts. Extensive experiments on large-scale datasets (Market-1501 and DukeMTMC-reID) demonstrate the superior performance of MEB-Net over the state-of-the-arts.