3D traffic scene comprises various 3D information about car objects, including their pose and shape. However, most recent studies pay relatively less attention to reconstructing detailed shapes. Furthermore, most of them treat each 3D object as an independent one, resulting in losses of relative context inter-objects and scene context reflecting road circumstances. A novel monocular 3D pose and shape reconstruction algorithm, based on bi-contextual attention and attention-guided modeling (BAAM), is proposed in this work. First, given 2D primitives, we reconstruct 3D object shape based on attention-guided modeling that considers the relevance between detected objects and vehicle shape priors. Next, we estimate 3D object pose through bi-contextual attention, which leverages relation-context inter objects and scene-context between an object and road environment. Finally, we propose a 3D nonmaximum suppression algorithm to eliminate spurious objects based on their Bird-Eye-View distance. Extensive experiments demonstrate that the proposed BAAM yields state-of-the-art performance on ApolloCar3D. Also, they show that the proposed BAAM can be plugged into any mature monocular 3D object detector on KITTI and significantly boost their performance. Code is available at https://github.com/gywns6287/BAAM.

Although 3D perception for autonomous vehicles has focused on frontal-view information, more than half of fatal accidents occur due to side impacts in practice (e.g., T-bone crash). Motivated by this fact, we investigate the problem of side-view depth estimation, especially for monocular fisheye cameras, which provide wide FoV information. However, since fisheye cameras head road areas, it observes road areas mostly and results in severe distortion on object areas, such as vehicles or pedestrians. To alleviate these issues, we propose a new fisheye depth estimation network, SlaBins, that infers an accurate and dense depth map based on a geometric property of road environments; most objects are standing (i.e., orthogonal) on the road environments. Concretely, we introduce a slanted multi-cylindrical image (MCI) representation, which allows us to describe a distance as a radius to a cylindrical layer orthogonal to the ground regardless of the camera viewing direction. Based on the slanted MCI, we estimate a set of adaptive bins and a per-pixel probability map for depth estimation. Then by combining it with the estimated slanted angle of viewing direction, we directly infer a dense and accurate depth map for fisheye cameras. Experiments demonstrate that SlaBins outperforms the state-of-the-art methods in both qualitative and quantitative evaluation on the SynWoodScape and KITTI-360 depth datasets. For more information, you can visit our project page https://syniez.github.io/SlaBins/.

This paper presents a new method for generating speed profiles for autonomous vehicles using a Transformer-based network that predicts the coefficients of quintic polynomials. To train and validate the network, we curate a dataset of 500K simulated urban driving scenarios, where the ground truths are obtained by running offline model predictive control (MPC) optimization. We also present tailored loss functions to emulate MPC behavior and constrain upper-and-lower boundary conditions to provide feasible speed profiles. Extensive experimental results demonstrate the efficacy of the proposed method in providing high-quality speed profiles for a large number of path candidates and long planning horizons. The proposed method is capable of generating efficient speed profiles for 1024 path candidates within 30ms.

A novel algorithm to detect road lanes in the eigen-lane space is proposed in this paper. First, we introduce the notion of eigenlanes, which are data-driven descriptors for structurally diverse lanes, including curved, as well as straight, lanes. To obtain eigenlanes, we perform the best rank-M approximation of a lane matrix containing all lanes in a training set. Second, we generate a set of lane candi-dates by clustering the training lanes in the eigenlane space. Third, using the lane candidates, we determine an optimal set of lanes by developing an anchor-based detection net-work, called SIIC-Net. Experimental results demonstrate that the proposed algorithm provides excellent detection performance for structurally diverse lanes. Our codes are available at https://github.com/dongkwonjin/Eigenlanes.

We present a self-supervised depth estimation approach using a unified volumetric feature fusion for surround-view images. Given a set of surround-view images, our method constructs a volumetric feature map by extracting image feature maps from surround-view images and fuse the feature maps into a shared, unified 3D voxel space. The volumetric feature map then can be used for estimating a depth map at each surround view by projecting it into an image coordinate. A volumetric feature contains 3D information at its local voxel coordinate; thus our method can also synthesize a depth map at arbitrary rotated viewpoints by projecting the volumetric feature map into the target viewpoints. Furthermore, assuming static camera extrinsics in the multi-camera system, we propose to estimate a canonical camera motion from the volumetric feature map. Our method leverages 3D spatio- temporal context to learn metric-scale depth and the canonical camera motion in a self-supervised manner. Our method outperforms the prior arts on DDAD and nuScenes datasets, especially estimating more accurate metric-scale depth and consistent depth between neighboring views.

A novel algorithm to detect an optimal set of semantic lines is proposed in this work. We develop two networks: selection network (S-Net) and harmonization network (H-Net). First, S-Net computes the probabilities and offsets of line candidates. Second, we filter out irrelevant lines through a selection-and-removal process. Third, we construct a complete graph, whose edge weights are computed by H-Net. Finally, we determine a maximal weight clique representing an optimal set of semantic lines. Moreover, to assess the overall harmony of detected lines, we propose a novel metric, called HIoU. Experimental results demonstrate that the proposed algorithm can detect harmonious semantic lines effectively and efficiently. Our codes are available at https://github.com/dongkwonjin/Semantic-Line-MWCS.

Maps are an increasingly important tool in our daily lives, yet their rich semantic content still largely depends on manual input. Motivated by the broad availability of geo-tagged street-view images, we propose a new task aiming to make the map update process more proactive. We focus on automatically detecting changes of Points of Interest (POIs), specifically stores or shops of any kind, based on visual input. Faced with the lack of an appropriate benchmark, we build and release a large dataset, captured in two large shopping centers, that comprises 33K geo-localized images and 578 POIs. We then design a generic approach that compares two image sets captured in the same venue at different times and outputs POI changes as a ranked list of map locations. In contrast to logo or franchise recognition approaches, our system does not depend on an external franchise database. It is instead inspired by recent deep metric learning approaches that learn a similarity function fit to the task at hand. We compare various loss functions to learn a metric aligned with the POI change detection goal, and report promising results.

A novel algorithm to estimate instance-level future motion in a single image is proposed in this paper. We first represent the future motion of an instance with its direction, speed, and action classes. Then, we develop a deep neural network that exploits different levels of semantic information to perform the future motion estimation. For effective future motion classification, we adopt ordinal regression. Especially, we develop the cyclic ordinal regression scheme using binary classifiers. Experiments demonstrate that the proposed algorithm provides reliable performance and thus can be used effectively for vision applications, including single and multi object tracking. Furthermore, we release the future motion (FM) dataset, collected from diverse sources and annotated manually, as a benchmark for single-image future motion estimation.

Recent works on domain adaptation exploit adversarial training to obtain domain-invariant feature representations from the joint learning of feature extractor and domain discriminator networks. However, domain adversarial methods render suboptimal performances since they attempt to match the distributions among the domains without considering the task at hand. We propose Drop to Adapt (DTA), which leverages adversarial dropout to learn strongly discriminative features by enforcing the cluster assumption. Accordingly, we design objective functions to support robust domain adaptation. We demonstrate efficacy of the proposed method on various experiments and achieve consistent improvements in both image classification and semantic segmentation tasks. Our source code is available at https://github.com/postBG/DTA.pytorch.

In this paper, we tackle unsupervised domain adaptation, where a target domain is unlabeled and lies on a considerably different distribution from a source domain. To alleviate such data discrepancies, we coin a novel deep neural network architecture that consists of a classifier and a domain discriminator on top of a shared feature extractor. Toward efficient regularization, we delegate a generation of the adversarial attacks to the domain discriminator. We then leverage the domain adversarial images to let the classification network learn important semantic features across the domains. Specifically, we employ consistency loss function that enables the joint use of clean and adversarial data. We present extensive experimental results on various domain adaptation benchmarks to show the efficacy of the proposed method.

We propose an image based end-to-end learning framework that helps lane-change decisions for human drivers and autonomous vehicles. The proposed system, Safe Lane-Change Aid Network (SLCAN), trains a deep convolutional neural network to classify the status of adjacent lanes from rear view images acquired by cameras mounted on both sides of the vehicle. Rather than depending on any explicit object detection or tracking scheme, SLCAN reads the whole input image and directly decides whether initiation of the lane-change at the moment is safe or not. We collected and annotated 77,273 rear side view images to train and test SLCAN. Experimental results show that the proposed framework achieves 96.98% classification accuracy although the test images are from unseen roadways. We also visualize the saliency map to understand which part of image SLCAN looks at for correct decisions.

In this work, we propose a stochastic model for curvilinear structure reconstruction using morphological profiles of path opening operator. We apply the support vector machine classifier to obtain initial probabilities to belong to line network for each pixel. Then, we formulate a stochastic optimization problem that detects line segments corresponding to the latent curvilinear structure in a scene. Experimental results on DNA filament and remote sensing images validate the effectiveness of the proposed algorithm when compared to other recent methods.

We propose a new model for wrinkle detection in human faces using a marked point process. In order to detect an arbitrary shape of wrinkles, we represent them as a set of line segments, where each segment is characterized by its length and orientation. We propose a probability density of wrinkle model which exploits local edge profile and geometric properties of wrinkles. To optimize the probability density of wrinkle model, we employ reversible jump Markov chain Monte Carlo sampler with delayed rejection. Experimental results demonstrate that the new algorithm detects facial wrinkles more accurately than a recent state-of-the-art method.

A novel motion-compensated frame interpolation (MCFI) algorithm to increase video temporal resolutions based on multihypothesis motion estimation and texture optimization is proposed in this paper. Initially, we form multiple motion hypotheses for each pixel by employing different motion estimation parameters, i.e., different block sizes and directions. Then, we determine the best motion hypothesis for each pixel by solving a labeling problem and optimizing the parameters. In the labeling problem, the cost function is composed of color, shape, and smoothness terms. Finally, we refine the motion hypothesis field based on the texture optimization technique and blend multiple source pixels to interpolate each pixel in the intermediate frame. Simulation results demonstrate that the proposed algorithm provides significantly better MCFI performance than conventional algorithms.

A novel motion-compensated frame interpolation algorithm to increase video frame rates is proposed in this work, which employs texture optimization techniques to refine inaccurate motion vector fields. We enforce the congruent segment constraint in the motion refinement so that matching objects have similar local image structures. More specifically, we use the congruence energy as well as the appearance energy in the motion optimization to estimate high quality motion vectors. Simulation results show that, by preserving complex object shapes and texture, the proposed algorithm provides more faithful intermediate frames than conventional algorithms.

A novel algorithm for thumbnail generation, which preserves characteristic features of a source image including blurs and textures, is proposed in this work. When a source image is subsampled to generate a thumbnail, important visible cues, such as blurs and noises, are lost. To overcome this drawback, we first create multiple thumbnail candidates that accentuate three classes of image features: focal blur, motion blur, and detail. Then, we obtain the final thumbnail by composing these candidates adaptively. Assuming that image features are spatially varying but locally static, we formulate the composition task as a labeling problem, and employ the graph-cut optimization technique to solve the problem. Simulation results demonstrate that the proposed algorithm provides feature-preserving thumbnails efficiently.