🤖 AI Summary
To address the challenge of scene understanding in unstructured road environments—particularly those characteristic of Indian urban and rural areas—this paper introduces a high-difficulty Indian driving dataset and proposes a multi-level semantic segmentation framework. Methodologically, we systematically evaluate five mainstream architectures—U-Net, U-Net+ResNet50, DeepLabV3, PSPNet, and SegNet—using mean Intersection-over-Union (mIoU) as the unified metric for pixel-wise drivable area, obstacle, and roadside object segmentation. Our contributions are threefold: (1) we present the first benchmark dataset specifically designed for typical Indian unstructured scenarios; (2) through cross-architecture analysis, we reveal differential robustness under challenging conditions including variable illumination, absent road signage, and ambiguous road boundaries; and (3) our framework achieves the state-of-the-art mIoU of 0.6496 on this dataset, significantly enhancing fine-grained scene parsing capability in complex traffic environments.
📝 Abstract
Autonomous vehicles are the next revolution in the automobile industry and they are expected to revolutionize the future of transportation. Understanding the scenario in which the autonomous vehicle will operate is critical for its competent functioning. Deep Learning has played a massive role in the progress that has been made till date. Semantic Segmentation, the process of annotating every pixel of an image with an object class, is one crucial part of this scene comprehension using Deep Learning. It is especially useful in Autonomous Driving Research as it requires comprehension of drivable and non-drivable areas, roadside objects and the like. In this paper semantic segmentation has been performed on the Indian Driving Dataset which has been recently compiled on the urban and rural roads of Bengaluru and Hyderabad. This dataset is more challenging compared to other datasets like Cityscapes, since it is based on unstructured driving environments. It has a four level hierarchy and in this paper segmentation has been performed on the first level. Five different models have been trained and their performance has been compared using the Mean Intersection over Union. These are UNET, UNET+RESNET50, DeepLabsV3, PSPNet and SegNet. The highest MIOU of 0.6496 has been achieved. The paper discusses the dataset, exploratory data analysis, preparation, implementation of the five models and studies the performance and compares the results achieved in the process.