TICaM (Time-of-flight In-car Cabin Monitoring)

TICaM 是一个飞行时间车内监控数据集，用于使用单个广角深度相机进行车辆内部监控。该数据集解决了其他可用的车内数据集在标记类别范围、记录场景和提供注释方面的不足；同时。它包含一个详尽的驾驶时执行的动作列表和多模态标记图像（深度、RGB 和 IR），具有用于 2D 和 3D 对象检测、实例和语义分割以及 RGB 帧的活动注释的完整注释。除了真实记录之外，它还包含具有相同多模态图像和注释的车内图像合成数据集，为有效训练车厢监控系统和评估域适应方法提供了合成数据和真实数据的独特且极其有益的组合.

TICaM is a time-of-flight (ToF) in-vehicle monitoring dataset designed for vehicle interior surveillance using a single wide-angle depth camera. It addresses the shortcomings of other existing in-vehicle datasets in terms of labeled category scope, recorded scenarios and annotation provision. It contains an exhaustive list of actions performed while driving, as well as multimodally annotated images (depth, RGB and IR), with complete annotations for 2D and 3D object detection, instance and semantic segmentation, and activity annotations for RGB frames. In addition to real-world recordings, it also includes an in-vehicle synthetic image dataset with the same multimodal images and annotations, providing a unique and extremely valuable combination of synthetic and real data for effectively training in-vehicle surveillance systems and evaluating domain adaptation methods.

搜集汇总

背景与挑战

背景概述

TICaM是一个用于车内监控的飞行时间数据集，提供多模态图像（深度、RGB和IR）和完整注释，支持2D/3D检测、分割及活动识别。它结合了真实记录和合成数据，以弥补现有数据集的不足，并促进车厢监控系统的训练和域适应方法评估。