Integrating multispectral data has been demonstrated to be an effective solution for illumination-invariant pedestrian detection, in particular, RGB and thermal images can provide complementary information to handle light variations. However, most of the current multispectral detectors fuse the multimodal features by simple concatenation, without discovering their latent relationships. In this paper, we propose a cross-modal feature learning (CFL) module, based on a split-and-aggregation strategy, to explicitly explore both the shared and modality-specific representations between paired RGB and thermal images. We insert the proposed CFL module into multiple layers of a two-branch-based pedestrian detection network, to learn the cross-modal representations in diverse semantic levels. By introducing a segmentation-based auxiliary task, the multimodal network is trained end-to-end by jointly optimizing a multi-task loss. On the other hand, to alleviate the reliance of existing multispectral pedestrian detectors on thermal images, we propose a knowledge distillation framework to train a student detector, which only receives RGB images as input and distills the cross-modal representations guided by a well-trained multimodal teacher detector. In order to facilitate the cross-modal knowledge distillation, we design different distillation loss functions for the feature, detection and segmentation levels. Experimental results on the public KAIST multispectral pedestrian benchmark validate that the proposed cross-modal representation learning and distillation method achieves robust performance.

Invertible image decolorization is a useful color compression technique to reduce the cost in multimedia systems. Invertible decolorization aims to synthesize faithful grayscales from color images, which can be fully restored to the original color version. In this paper, we propose a novel color compression method to produce invertible grayscale images using invertible neural networks (INNs). Our key idea is to separate the color information from color images, and encode the color information into a set of Gaussian distributed latent variables via INNs. By this means, we force the color information lost in grayscale generation to be independent of the input color image. Therefore, the original color version can be efficiently recovered by randomly re-sampling a new set of Gaussian distributed variables, together with the synthetic grayscale, through the reverse mapping of INNs. To effectively learn the invertible grayscale, we introduce the wavelet transformation into a UNet-like INN architecture, and further present a quantization embedding to prevent the information omission in format conversion, which improves the generalizability of the framework in real-world scenarios. Extensive experiments on three widely used benchmarks demonstrate that the proposed method achieves a state-of-the-art performance in terms of both qualitative and quantitative results, which shows its superiority in multimedia communication and storage systems.

Facial expression recognition (FER) is of great interest to the current studies of human-computer interaction. In this paper, we propose a novel geometry-guided facial expression recognition framework, based on graph convolutional networks and transformers, to perform effective emotion recognition from videos. Specifically, we detect and utilize facial landmarks to construct a spatial-temporal graph, based on both the landmark coordinates and local appearance, for representing a facial expression sequence. The graph convolutional blocks and transformer modules are employed to produce high-semantic emotion-related representations from the structured facial graphs, which facilitate the framework to establish both the local and non-local dependency between the vertices. Moreover, spatial and temporal attention mechanisms are introduced into graph-based learning to promote FER reasoning, via the emphasis on the most informative facial components and frames. Extensive experiments demonstrate that the proposed framework achieves promising performance for geometry-based FER and shows great generalization and robustness in real-world applications.

Learning discriminative representations with good robustness from facial observations serves as a fundamental step towards intelligent facial expression recognition (FER). In this paper, we propose a novel geometry-aware FER framework to boost the FER performance based on both the geometric and appearance knowledge. Specifically, we propose an encoding strategy for facial landmarks, and adopt a graph convolutional network (GCN) to fully explore the structural information of the facial components behind different expressions. A convolutional neural network (CNN) is further applied to the whole facial observation to learn the global characteristics of different expressions. The features from these two networks are fused into a comprehensive high-semantic representation, which promotes the FER reasoning from both visual and structural perspectives. Moreover, to facilitate the networks to concentrate on the most informative facial regions and components, we introduce multi-level attention mechanisms into the proposed framework, which enhance the reliability of the learned representations for effective FER. Experiments on two challenging FER benchmarks demonstrate that the attentive graph-based learning on the facial geometry boosts the FER accuracy. Furthermore, the insensitivity of the geometric information to the appearance variations also improves the generalization of the proposed framework.

Volume Projection Imaging from ultrasound data is a promising technique to visualize spine features and diagnose Adolescent Idiopathic Scoliosis. In this paper, we present a novel multi-task framework to reduce the scan noise in volume projection images and to segment different spine features simultaneously, which provides an appealing alternative for intelligent scoliosis assessment in clinical applications. Our proposed framework consists of two streams: i) A noise removal stream based on generative adversarial networks, which aims to achieve effective scan noise removal in a weakly-supervised manner, i.e., without paired noisy-clean samples for learning; ii) A spine segmentation stream, which aims to predict accurate bone masks. To establish the interaction between these two tasks, we propose a selective feature-sharing strategy to transfer only the beneficial features, while filtering out the useless or harmful information. We evaluate our proposed framework on both scan noise removal and spine segmentation tasks. The experimental results demonstrate that our proposed method achieves promising performance on both tasks, which provides an appealing approach to facilitating clinical diagnosis.

In medical image analysis, anatomical landmarks usually contain strong prior knowledge of their structural information. In this paper, we propose to promote medical landmark localization by modeling the underlying landmark distribution via normalizing flows. Specifically, we introduce the flow-based landmark distribution prior as a learnable objective function into a regression-based landmark localization framework. Moreover, we employ an integral operation to make the mapping from heatmaps to coordinates differentiable to further enhance heatmap-based localization with the learned distribution prior. Our proposed Normalizing Flow-based Distribution Prior (NFDP) employs a straightforward backbone and non-problem-tailored architecture (i.e., ResNet18), which delivers high-fidelity outputs across three X-ray-based landmark localization datasets. Remarkably, the proposed NFDP can do the job with minimal additional computational burden as the normalizing flows module is detached from the framework on inferencing. As compared to existing techniques, our proposed NFDP provides a superior balance between prediction accuracy and inference speed, making it a highly efficient and effective approach. The source code of this paper is available at https://github.com/jacksonhzx95/NFDP.