This work investigates the functionality of Semantic information in Contrastive Learning (SemCL). An advanced pretext task is designed: a contrast is performed between each object and its environment, taken from a scene. This allows the SemCL pretrained model to extract objects from their environment in an image, significantly improving the spatial understanding of the pretrained models. Downstream tasks of semantic/instance segmentation, object detection and depth estimation are implemented on PASCAl VOC, Cityscapes, COCO, KITTI, etc. SemCL pretrained models substantially outperform ImageNet pretrained counterparts and are competitive with well-known works on downstream tasks. The results suggest that a dedicated pretext task lever-aging semantic information can be powerful in benchmarks related to spatial understanding. The code is available at https://github.com/sjiang95/semcl.

We present One-Shot Affordance Learning (OSAL): a unified pipeline that learns manipulation for articulated objects by observing human demonstration only once. The key idea of our method is to embody affordance of articulated objects with an open-loop trajectory conditioned on a certain area of the object's surface. It serves as a simplified object-centric manipulation representation, which can be easily transferred into robot motion, while traditional methods fail to deal with the configuration difference between human hands and robot end effectors. Our system extracts the embodied affordance by focusing on hand action's effect on the object, and further grounds such affordance into object visual features through self-supervised learning for novel object configurations. We evaluated our method on a collection of real-life objects and furniture and demonstrated high success rates. With our system, humans only need to manipulate a novel object once with any gesture to transfer that manipulation skill to the robot, which we believe to be a highly efficient and user-friendly paradigm oriented for future real-life robots.

The tradeoff between the field-of-view and resolution of conventional onboard vision systems primarily results from their fixed optical components. We propose a novel active vision system, Falcon, as an optimal solution. This system comprises an electric zoom lens connected to a high-speed camera with a pair of galvanometer mirrors, enabling high-resolution imaging of a moving object across a wide range, from near to far. To ensure accurate calibration of the Falcon system, we introduce a novel mapping-based calibration method using external cameras. We also present a robust and lightweight visual feedback method that utilizes this mapping-based calibration for effective object tracking. The effectiveness of the Falcon system is verified by constructing a prototype and conducting tracking experiments in an indoor setting, which demonstrated the superior performance of our method. Additionally, we successfully achieved continuous and high-resolution imaging of a curved mirror on public roads while the vehicle was moving.

We propose an accurate and robust inter-vehicle distance estimation method using highspeed stereo vision. The framework involves two phases: a tracking phase, wherein a preceding vehicle is accurately and stably tracked by a tracking algorithm optimized for stereo high-speed vision, and a distance estimation phase, wherein the inter-vehicle distance is estimated via a highly accurate scale estimation and aggregation method for multiple scale-based distance estimations to ensure that it is more accurate and robust without introducing a delay. Further, we propose patch multiplexing to realize accurate and efficient aggregation even in situations where the scale changes rapidly (e.g., emergency braking). Through comparative analysis using three real-world scenarios, we verify that the accuracy of inter-vehicle distance estimation using our approach is comparable to that of laser rangefinders. We also demonstrate that differential quantities, such as velocity and acceleration, could be accurately estimated using an adaptive Kalman filter. Our results will help develop safe and accurate truck platooning and adaptive cruise control systems.

We propose a driving safety support system (DSSS) that employs a high-speed vision system installed in the environment surrounding, for instance, highways, urban roads, and intersections. The aim of the system is to recognize potentially dangerous traffic situations, including those that are undetectable from a moving vehicle, and to use this information for supporting safe driving. The system consists of a vision network of synchronized high-speed cameras that are capable of acquiring images at one-millisecond intervals, and vehicles that are capable of communicating with this network through communication hubs. We conducted collision avoidance experiments and demonstrated that, by introducing high-speed vision, the proposed system can resolve the issue of slow reaction time, which is common to environmental vision systems.

This paper presents an approach for document rectification using a quasi-isometric mapping derived from a novel model that represents a deformed document surface. The model is composed of rectangle fragments of the developed document plane. This was realized by introducing gap relaxation which permits unconnected fragments. Our experiments show that our rectification approach can be applied to different types of document deformation.

The authors describe the main features of two recently established joint research laboratories for optoelectronics in Japan: the Optoelectronics Joint Research Laboratory, which existed from 1981 to 1986, and the Optoelectronics Technology Research Laboratory, which was established in 1986. The laboratories were established to make technical breakthroughs in materials technologies for future optoelectronic devices, specifically the optoelectronic integrated circuit. The background, organization, purpose, activities, and merits of this kind of joint research activity are discussed. In such laboratories, researchers on leave from member companies work together toward the same objective within a specific time period. The first trial of such a joint venture in Japan, in the late 1970s, was in the field of silicon large-scale integrated circuits. The change of emphasis from shorter-term to longer-term targets reflects the growth of the semiconductor electronics industry in Japan, but also results from experience with such an unusual form of joint laboratory. It is concluded that a joint laboratory works most effectively when new technological concepts are being explored, which may open up future markets. Focused in this way, a joint laboratory could lead the industry in those aspects of technology that precede competitive device development.>