Guide dog robots offer promising solutions to enhance mobility and safety for visually impaired individuals, addressing the limitations of traditional guide dogs, particularly in perceptual intelligence and communication. With the emergence of Vision-Language Models (VLMs), robots are now capable of generating natural language descriptions of their surroundings, aiding in safer decision-making. However, existing VLMs often struggle to accurately interpret and convey spatial relationships, which is crucial for navigation in complex environments such as street crossings. We introduce the SpaceAware Instruction Tuning (SAIT) dataset and the Space-Aware Benchmark (SA-Bench) to address the limitations of current VLMs in understanding physical environments. Our automated data generation pipeline focuses on the virtual path to the destination in 3D space and the surroundings, enhancing environmental comprehension and enabling VLMs to provide more accurate guidance to visually impaired individuals. We also propose an evaluation protocol to assess VLM effectiveness in delivering walking guidance. Comparative experiments demonstrate that our space-aware instruction-tuned model outperforms state-of-the-art algorithms. We have fully opensourced the SAIT dataset and SA-Bench, along with the related code, at https://github.com/byungokhan/Space-awareVLM.

Recent complicated problems require large-scale datasets and complex model architectures, however, it is difficult to train such large networks due to high computational issues. Significant efforts have been made to make the training more efficient such as momentum, learning rate scheduling, weight regularization, and meta-learning. Based on our observations on 1) high correlation between past eights and future weights, 2) conditions for beneficial weight prediction, and 3) feasibility of weight prediction, we propose a more general framework by intermittently skipping a handful of epochs by periodically forecasting near future weights, i.e., a Weight Nowcaster Network (WNN). As an add-on module, WNN predicts the future weights to make the learning process faster regardless of tasks and architectures. Experimental results show that WNN can significantly save actual time cost for training with an additional marginal time to train WNN. We validate the generalization capability of WNN under various tasks, and demonstrate that it works well even for unseen tasks. The code and pre-trained model are available at https://github.com/jjh6297/WNN.

Facial expressions are one of the most essential channels to communicate a person’s emotional state. In social interaction, the capability to accurately read subtle changes in facial expressions, which reveal emotional fluctuations, is critical for 1) comprehending others’ emotions in context and background situations, 2) identifying responsiveness to others’ emotions, and 3) developing social skills in human–computer interaction. In this paper, we first introduce automatic emotion change detection via facial expression that discovers timings or temporal locations in a video where facial expression significantly changes. We propose a weakly-supervised deep emotion change detection framework that does not require facial expression videos with expensive temporal annotations and instead learns static images for training. Incorporating these ideas, we performed extensive experiments to demonstrate fundamental insights into emotion change detection and the efficacy of our framework using three video datasets, i.e., CASME II, MMI, and our YoutubeECD. Furthermore, we modified our framework for temporal spotting, which is the most similar task to emotion change detection, and showed comparable results with state-of-the-art methods on CAS(ME)2, proving justification for the problem. Even though we only employed the AffectNet to train our framework rather than the CASME II, MMI, YoutubeECD, and CAS(ME)2, experimental results demonstrate its exceptional generalization capability in cross-dataset environments.

Conventional studies dealing with repetition detection have mainly focused on tasks of temporal localization or counting the number of repetitions in videos. However, direct discrimination between repetitive and non-repetitive behaviors in videos, called repetitive behavior classification (RBC), has attracted less attention despite its great potential and advantages of: 1) filling the demands in the fields such as classification of repetitive behaviors in children with autism spectrum disorder (ASD) and helping to alleviate manual and time-consuming diagnostic procedures, 2) directly learning representation of differences between repetition and non-repetition patterns along the temporal dimension, and 3) being an effective alternative to the existing repetition counting and temporal segmentation tasks that are struggling with insufficient data and laborious manual annotation effort. In this paper, to the best of our knowledge, we firstly cast the problem of the RBC using deep learning frameworks. For this, we propose a simple yet effective add-on network, SiRepNet, that exploits the Siamese network structure to learn the inherent properties of repetitive behaviors. We also composed the RBC dataset by re-purposing and re-organizing Kinetics and Countix datasets for training our method. To validate our ideas, we carried out extensive experiments on RBC datasets, which showed performance improvement over state-of-the-art video classification algorithms by simply attaching our scheme to them for the RBC task.

In this paper, a new mobile Augmented Reality (AR) framework which is scalable to the number of objects being augmented is proposed. The scalability is achieved by a visual word recognition module on the remote server and a mobile phone which detects, tracks, and augments target objects with the received information from the server. The server and the mobile phone are connected through a conventional Wi-Fi. In the experiment, it takes 0.2 seconds for the cold start of an AR service initiation on a 10k object database, which is fairly acceptable in a real-world AR application.