In this paper, we propose a convolutional neural network for thermal image enhancement by incorporating the brightness domain with a residual-learning technique, which improves the performance of enhancement and speed of convergence. Typically, the training domain uses the same domain as that of the target image; however, we evaluated several domains to determine the most suitable one for the network. In the analyses, we first compared the performance of networks that were trained by the corresponding regions of color-based and aligned infrared-based images, respectively, including thermal, far, and near spectra. Then, four RGB-based domains, namely, gray, lightness, intensity, and brightness were evaluated. Finally, the proposed network architecture was determined by considering the residual and brightness domains. The results of the analyses indicated that the brightness domain was the best training domain for enhancing the thermal images. The experimental results confirm that the proposed network, which can be trained in approximately one hour, outperforms the conventional learning-based approaches for thermal image enhancement, in terms of several image quality metrics and a qualitative evaluation. Furthermore, the results demonstrate that the brightness domain is effective as the training domain and can be used to increase the performance of existing networks.

Heterogeneous Face Recognition (HFR) is a task that matches faces across two different domains such as visible light (VIS), near-infrared (NIR), or the sketch domain. Due to the lack of databases, HFR methods usually exploit the pre-trained features on a large-scale visual database that contain general facial information. However, these pre-trained features cause performance degradation due to the texture discrepancy with the visual domain. With this motivation, we propose a graph-structured module called Relational Graph Module (RGM) that extracts global relational information in addition to general facial features. Because each identity's relational information between intra-facial parts is similar in any modality, the modeling relationship between features can help cross-domain matching. Through the RGM, relation propagation diminishes texture dependency without losing its advantages from the pre-trained features. Furthermore, the RGM captures global facial geometrics from locally correlated convolutional features to identify long-range relationships. In addition, we propose a Node Attention Unit (NAU) that performs node-wise recalibration to concentrate on the more informative nodes arising from relation-based propagation. Furthermore, we suggest a novel conditional-margin loss function ($C$ -softmax) for the efficient projection learning of the embedding vector in HFR. The proposed method outperforms other state-of-the-art methods on five HFR databases. Furthermore, we demonstrate performance improvement on three backbones because our module can be plugged into any pre-trained face recognition backbone to overcome the limitations of a small HFR database.

Instance segmentation has gained attention in various computer vision fields, such as autonomous driving, drone control, and sports analysis. Recently, many successful models have been developed, which can be classified into two categories: accuracy- and speed-focused. Accuracy and inference time are important for real-time applications of this task. However, these models just present inference time measured on different hardware, which makes their comparison difficult. This study is the first to evaluate and compare the performances of state-of-the-art instance segmentation models by focusing on their inference time in a fixed experimental environment. For precise comparison, the test hardware and environment should be identical; hence, we present the accuracy and speed of the models in a fixed hardware environment for quantitative and qualitative analyses. Although speed-focused models run in real-time on high-end GPUs, there is a trade-off between speed and accuracy when the computing power is insufficient. The experimental results show that a feature pyramid network structure may be considered when designing a real-time model, and a balance between the speed and accuracy must be achieved for real-time application.

Depth completion is the task of reconstructing dense depth images from sparse LiDAR data. LiDAR depth completion, for which LiDAR data is the only input, is an ill-posed and challenging problem owing to the underlying properties of LiDAR data: extremely few points, presence of discontinuities, and absence of texture information. Accordingly, most approaches are heavily dependent on guided color images, which leads to unsatisfactory results when the color images are degraded. To alleviate the dependency on color images but leverage this information during training, we present a deep convolutional neural network (CNN) consisting of depth and edge CNNs via transferring of knowledge. In order to compensate for the limitations of LiDAR data, we design the edge CNN to learn a gradient depth image from a powerful teacher network through the Knowledge-Distillation method. Since the teacher network is trained with color images, color-embedded information can be obtained in the test phase even if color images are not used as an input. We further propose a Self-Distillation method for transferring the color-embedded features from the edge CNN to the depth CNN. Enforcing the depth features to contain edge information hardly observed in LiDAR data enables the depth CNN to generate more edge-attentive and structure-preserving results. Our novel methods show remarkable results in outdoor and indoor environments for KITTI and NYU-Depth-V2 datasets. Experiments performed with low-channel LiDAR data in KITTI and few depth points in the NYU-Depth-V2 dataset show that our method is robust to data sparsity and applicable in various scenarios.

Studies on depth images containing three-dimensional information have been performed for many practical applications. However, the depth images acquired from depth sensors have inherent problems, such as missing values and noisy boundaries. These problems significantly affect the performance of applications that use a depth image as their input. This paper describes a depth enhancement algorithm based on a combination of color and depth information. To fill depth holes and recover object shapes, asynchronous cellular automata with neighborhood distance maps are used. Image segmentation and a weighted linear combination of spatial filtering algorithms are applied to extract object regions and fill disocclusion in the object regions. Experimental results on both real-world and public datasets show that the proposed method enhances the quality of the depth image with low computational complexity, outperforming conventional methods on a number of metrics. Furthermore, to verify the performance of the proposed method, we present stereoscopic images generated by the enhanced depth image to illustrate the improvement in quality.

Estimating frame-to-frame (F2F) visual odometry with monocular images has significant problems of propagated accumulated drift. We propose a learning-based approach for F2F monocular visual odometry estimation with novel and simple methods that consider the coherence of camera trajectories without any post-processing. The proposed network consists of two stages: initial estimation and error relaxation. In the first stage, the network learns disparity images to extract features and predicts relative camera pose between adjacent two frames through the attention, rotation, and translation networks. Then, loss functions are proposed in the error relaxation stage to reduce the local drift, increasing consistency under dynamic driving scenes. Moreover, our skip-ordering scheme shows the effectiveness of dealing with sequential data. Experiments with the KITTI benchmark dataset show that our proposed network outperforms other approaches with higher and more stable performance.