IDD
收藏资源简介:
IDD数据集由印度信息技术学院海得拉巴分校创建,专注于无约束环境下的自动驾驶导航问题。该数据集包含10,004张图像,精细标注了34个类别,源自印度道路的182个驾驶序列。数据集的创建旨在解决现有数据集在结构化环境中的局限性,提供更丰富的类别和更高的类别内多样性。IDD数据集特别适用于领域适应、少样本学习和行为预测等高级研究问题。
The IDD dataset, developed by the International Institute of Information Technology, Hyderabad (IIIT Hyderabad), focuses on the problem of autonomous driving navigation in unconstrained real-world environments. It contains 10,004 images with fine-grained annotations across 34 categories, sourced from 182 driving sequences collected on Indian public roads. The dataset was created to address the limitations of existing autonomous driving datasets that mostly target structured environments, by providing a richer set of categories and higher intra-class diversity. Notably, the IDD dataset is particularly applicable to advanced research topics such as domain adaptation, few-shot learning, and behavior prediction.

- IDD数据集首次发表,由印度理工学院马德拉斯分校的研究团队发布,旨在解决印度道路场景中的自动驾驶问题。
- IDD数据集首次应用于自动驾驶领域的研究,特别是在印度复杂道路环境下的目标检测和场景理解任务中。
- IDD数据集被广泛用于多个国际会议和期刊的论文中,成为评估和比较不同算法在印度道路场景中性能的标准数据集之一。
- IDD数据集的扩展版本发布,增加了更多的场景和标注信息,以支持更复杂的自动驾驶研究和应用。
- 1Indian Driving Dataset: A Benchmark for Vision-Based Driving Scene UnderstandingIndian Institute of Technology Hyderabad · 2019年
- 2A Comprehensive Analysis of Vision-Based Driving Scene Understanding Using the Indian Driving DatasetIndian Institute of Technology Bombay · 2020年
- 3Multi-Task Learning for Scene Understanding in Indian Driving DatasetIndian Institute of Technology Madras · 2021年
- 4Semantic Segmentation on Indian Driving Dataset Using Deep Learning TechniquesIndian Institute of Technology Delhi · 2022年



