Most existing abandoned object detection algorithms use foreground information generated from background models. Detection using the background subtraction technique performs well under normal circumstances. However, it has a significant problem where the foreground information is gradually absorbed into the background as time passes and disappears, making it very vulnerable to sudden illumination changes that increase the false alarm rate. This paper presents an algorithm for detecting abandoned objects using a dual background model, which is robust even in illumination changes as well as other complex circumstances like occlusion, long-term abandonment, and owner re-attendance. The proposed algorithm can adapt quickly to various illumination changes. And also, it can precisely track the target objects to determine whether it is abandoned regardless of the existence of foreground information and the effect from the illumination changes, thanks to the largest-contour-based presence authentication mechanism proposed in this paper. For performance evaluation, we trialed the algorithm with the PETS2006, ABODA datasets as well as our dataset, especially to demonstrate its robustness in various illumination changes.

This paper introduces a novel approach for self-supervised monocular depth estimation. The model is trained on stereo–image (left–right pair) data and incorporates carefully designed perceptual image quality assessment-based loss functions for image reconstruction and left–right image difference. The fidelity of the reconstructed images, obtained by warping the input images using the predicted disparity maps, significantly influences the accuracy of depth estimation in self-supervised monocular depth networks. The suggested LPIPS (Learned Perceptual Image Patch Similarity)-based evaluation of image reconstruction accurately emulates human perceptual mechanisms to quantify the quality of reconstructed images, serving as an image reconstruction loss. Consequently, it facilitates the gradual convergence of the reconstructed images toward a greater similarity with the target images during the training process. Stereo–image pair often exhibits slight discrepancies in brightness, contrast, color, and camera angle due to factors like lighting conditions and camera calibration inaccuracies. These factors limit the improvement of image reconstruction quality. To address this, the left–right difference image loss is introduced, aimed at aligning the disparities between the actual left–right image pair and the reconstructed left–right image pair. Due to the tendency of distant pixel values to approach zero in the difference images derived from the left and right source images of stereo pairs, this loss progressively steers the distant pixel values of the reconstructed difference images toward a convergence with zero. Hence, the use of this loss has demonstrated its efficacy in mitigating distortions in distant regions while enhancing overall performance. The primary objective of this study is to introduce and validate the effectiveness of LPIPS-based image reconstruction and left–right difference image losses in the context of monocular depth estimation. To this end, the proposed loss functions have been seamlessly integrated into a straightforward single-task stereo–image learning framework, incorporating simple hyperparameters. Notably, our approach achieves superior results compared to other state-of-the-art methods, even those adopting more intricate hybrid data and multi-task learning strategies.

This article proposes a novel approach for predicting depth from a single image in a self-supervised manner, specifically using generative adversarial networks (GANs) with enhancements to improve training performance. We train our model to generate disparity maps or inverse depth maps using left/right stereo image supervision rather than direct supervision from depth ground truth. We avoid the problem of training instability in standard GANs by employing a relativistic discriminator that evaluates how realistic an input image is. Since the reconstructed image quality determines the disparity map’s accuracy, we adopt the GMSD, one of the standout IQA algorithms, as a part of the image reconstruction loss function. A VGG-19 pre-trained model-based perceptual loss is used to optimize our model in a feature space instead of pixel space and to enhance reconstructed images’ perceptual quality with the learning-based perspective. Also, we add another symmetric generator that takes the right images as input, which helps to maintain the consistency of left/right views and enrich our training data with reasonable overhead on the training time. Our network is trained to predict depths of $640 \times 192$ images using the KITTI stereo dataset, which shows an outstanding performance compared to state-of-the-art works in various evaluation metrics. We verify our method’s modules by performing an ablation study and evaluate the generalization performance on CityScapes and Make3D datasets.

In this paper, we propose a novel unsupervised learning-based model for estimating the depth of monocular images by integrating a simple ResNet-based auto-encoder and some special loss functions. We use only stereo images obtained from binocular cameras as training data without using depth ground-truth data. Our model basically outputs a disparity map that is necessary to warp an input image to an image corresponding to a different viewpoint. When the input image is warped using the output-disparity map, distortions of various patterns inevitably occur in the reconstructed image. During the training process, the occurrence frequency and size of these distortions gradually decrease, while the similarity between the reconstructed and target images increases, which proves that the accuracy of the predicted disparity maps also increases. Therefore, one of the important factors in this type of training is an efficient loss function that accurately measures how much the difference in quality between the reconstructed and target images is and guides the gap to be properly and quickly closed as the training progresses. In recent related studies, the photometric difference was calculated through simple methods such as L1 and L2 loss or by combining one of these with a traditional computer vision-based hand-coded image-quality assessment algorithm such as SSIM. However, these methods have limitations in modeling various patterns at the level of the human visual system. Therefore, the proposed model uses a pre-trained perceptual image-quality assessment model that effectively mimics human-perception mechanisms to measure the quality of distorted images as image-reconstruction loss. In order to highlight the performance of the proposed loss functions, a simple ResNet50-based network is adopted in our model. We trained our model using stereo images of the KITTI 2015 driving dataset to measure the pixel-level depth for 768 × 384 images. Despite the simplicity of the network structure, thanks to the effectiveness of the proposed image-reconstruction loss, our model outperformed other state-of-the-art studies that have been trained in unsupervised methods on a variety of evaluation indicators.