Progressive Query Refinement Framework for Bird's-Eye-View Semantic Segmentation from Surrounding Images

Expressing images with Multi-Resolution (MR) features has been widely adopted in many computer vision tasks. In this paper, we introduce the MR concept into Bird's-Eye-View (BEV) semantic segmentation for autonomous driving. This introduction enhances our model's ability to capture both global and local characteristics of driving scenes through our proposed residual learning. Specifically, given a set of MR BEV query maps, the lowest resolution query map is initially updated using a View Transformation (VT) encoder. This updated query map is then upscaled and merged with a higher resolution query map to undergo further updates in a subsequent VT encoder. This process is repeated until the resolution of the updated query map reaches the target. Finally, the lowest resolution map is added to the target resolution to generate the final query map. During training, we enforce both the lowest and final query maps to align with the ground-truth BEV semantic map to help our model effectively capture the global and local characteristics. We also propose a visual feature interaction network that promotes interactions between features across images and across feature levels, thus highly contributing to the performance improvement. We evaluate our model on a large-scale real-world dataset. The experimental results show that our model outperforms the SOTA models in terms of IoU metric. Codes are available at https://github.com/d1024choi/ProgressiveQueryRefineNet

Driving Experience Transfer Method for End-to-End Control of Self-Driving Cars

In this paper, we present a transfer learning method for the end-to-end control of self-driving cars, which enables a convolutional neural network (CNN) trained on a source domain to be utilized for the same task in a different target domain. A conventional CNN for the end-to-end control is designed to map a single front-facing camera image to a steering command. To enable the transfer learning, we let the CNN produce not only a steering command but also a lane departure level (LDL) by adding a new task module, which takes the output of the last convolutional layer as input. The CNN trained on the source domain, called source network, is then utilized to train another task module called target network, which also takes the output of the last convolutional layer of the source network and is trained to produce a steering command for the target domain. The steering commands from the source and target network are finally merged according to the LDL and the merged command is utilized for controlling a car in the target domain. To demonstrate the effectiveness of the proposed method, we utilized two simulators, TORCS and GTAV, for the source and the target domains, respectively. Experimental results show that the proposed method outperforms other baseline methods in terms of stable and safe control of cars.