An official website of the United States government
After the 2017 TuSimple Lane Detection Challenge, its dataset and evaluation based on accuracy and F1 score have become the de facto standard to measure the performance of lane detection methods. While they have played a major role in improving the performance of lane detection methods, the validity of this evaluation method in downstream tasks has not been adequately researched. In this study, we design 2 new driving-oriented metrics for lane detection: End-to-End Lateral Deviation metric (E2E-LD) is directly formulated based on the requirements of autonomous driving, a core downstream task of lane detection; Per-frame Simulated Lateral Deviation metric (PSLD) is a lightweight surrogate metric of E2E-LD. To evaluate the validity of the metrics, we conduct a large-scale empirical study with 4 major types of lane detection approaches on the TuSimple dataset and our newly constructed dataset Comma2k19-LD. Our results show that the conventional metrics have strongly negative correlations (≤-0.55) with E2E-LD, meaning that some recent improvements purely targeting the conventional metrics may not have led to meaningful improvements in autonomous driving, but rather may actually have made it worse by overfitting to the conventional metrics. As autonomous driving is a security/safety-critical system, the underestimation of robustness hinders the sound development of practical lane detection models. We hope that our study will help the community achieve more downstream task-aware evaluations for lane detection.
more »
« less
A Deep Learning Approach for Lane Detection
State-of-the-art lane detection methods use a variety of deep learning techniques for lane feature extraction and prediction, demonstrating better performance than conventional lane detectors. However, deep learning approaches are computationally demanding and often fail to meet real-time requirements of autonomous vehicles. This paper proposes a lane detection method using a light-weight convolutional neural network model as a feature extractor exploiting the potential of deep learning while meeting real-time needs. The developed model is trained with a dataset containing small image patches of dimension 16 × 64 pixels and a non-overlapping sliding window approach is employed to achieve fast inference. Then, the predictions are clustered and fitted with a polynomial to model the lane boundaries. The proposed method was tested on the KITTI and Caltech datasets and demonstrated an acceptable performance. We also integrated the detector into the localization and planning system of our autonomous vehicle and runs at 28 fps in a CPU on image resolution of 768 × 1024 meeting real-time requirements needed for self-driving cars.
more »
« less
- PAR ID:
- 10326309
- Date Published:
- Journal Name:
- 2021 IEEE International Intelligent Transportation Systems Conference (ITSC)
- Page Range / eLocation ID:
- 1527 to 1532
- Format(s):
- Medium: X
- Sponsoring Org:
- National Science Foundation
More Like this
-
-
Vehicle to Vehicle (V2V) communication allows vehicles to wirelessly exchange information on the surrounding environment and enables cooperative perception. It helps prevent accidents, increase the safety of the passengers, and improve the traffic flow efficiency. However, these benefits can only come when the vehicles can communicate with each other in a fast and reliable manner. Therefore, we investigated two areas to improve the communication quality of V2V: First, using beamforming to increase the bandwidth of V2V communication by establishing accurate and stable collaborative beam connection between vehicles on the road; second, ensuring scalable transmission to decrease the amount of data to be transmitted, thus reduce the bandwidth requirements needed for collaborative perception of autonomous driving vehicles. Beamforming in V2V communication can be achieved by utilizing image-based and LIDAR’s 3D data-based vehicle detection and tracking. For vehicle detection and tracking simulation, we tested the Single Shot Multibox Detector deep learning-based object detection method that can achieve a mean Average Precision of 0.837 and the Kalman filter for tracking. For scalable transmission, we simulate the effect of varying pixel resolutions as well as different image compression techniques on the file size of data. Results show that without compression, the file size for only transmitting the bounding boxes containing detected object is up to 10 times less than the original file size. Similar results are also observed when the file is compressed by lossless and lossy compression to varying degrees. Based on these findings using existing databases, the impact of these compression methods and methods of effectively combining feature maps on the performance of object detection and tracking models will be further tested in the real-world autonomous driving system.more » « less
-
Optical coherence tomography (OCT) imaging enables high resolution visualization of sub-surface tissue microstructures. However, OCT image analysis using deep learning is hampered by limited diverse training data to meet performance requirements and high inference latency for real-time applications. To address these challenges, we developed Octascope, a lightweight domain-specific convolutional neural network (CNN) - based model designed for OCT image analysis. Octascope was pre-trained using a curriculum learning approach, which involves sequential training, first on natural images (ImageNet), then on OCT images from retinal, abdominal, and renal tissues, to progressively acquire transferable knowledge. This multi-domain pre-training enables Octascope to generalize across varied tissue types. In two downstream tasks, Octascope demonstrated notable improvements in predictive accuracy compared to alternative approaches. In the epidural tissue detection task, our method surpassed single-task learning with fine-tuning by 9.13% and OCT-specific transfer learning by 5.95% in accuracy. Octascope outperformed VGG16 and ResNet50 by 5.36% and 6.66% in a retinal diagnosis task, respectively. In comparison to a Transformer-based OCT foundation model - RETFound, Octascope delivered 2 to 4.4 times faster inference speed with slightly better predictive accuracies in both downstream tasks. Octascope represented a significant advancement for OCT image analysis by providing an effective balance between computational efficiency and diagnostic accuracy for real-time clinical applications.more » « less
-
Panoptic perception models in autonomous driving use deep learning models to interpret their surroundings and make real-time decisions. However, these models are susceptible, carefully designed noise can fool models all while being imperceptible to humans. In this work, we investigate the impact of blackbox adversarial noise attacks on three core perception tasks: drivable area recognition, lane line segmentation, and object detection. Unlike white-box attacks, black-box attacks assume no knowledge of the model’s internal parameters making them a more realistic and challenging threat scenario. Our goal is to evaluate how such an attack affects the model’s predictions and explore countermeasures towards such attacks. In response to our implemented attack, we have tested various defense methods. With each defense method, we have assessed the recovery on prediction accuracy. This research aims to provide valuable insights into the vulnerabilities of panoptic perception models and highlights strategies for enhancing their resilience against adversarial manipulation within real-world scenarios. All our attacks are performed against images from the BDD100K dataset.more » « less
-
This research develops, compares, and analyzes both a traditional algorithm using computer vision and a deep learning model to deal with dynamic road conditions. In the final testing, the deep learning model completed the target of five laps for both the inner and outer lane, whereas the computer vision algorithm only completed almost three laps for the inner lane and slightly over four laps for the outer. After conducting statistical analysis on the results of our deep learning model by finding the p-value between the absolute error and squared error of the self-driving algorithm in the outer lane and inner lane, we find that our results are statistically significant based on a two-tailed T test with unequal variances where the p-value for absolute error is 0.009, and 0.001 for squared error. Self-driving vehicles are not only complex, but they are growing in necessity—therefore, finding an optimal solution for lane detection in dynamic conditions is crucial to continue innovation.more » « less
- Free Publicly Accessible Full Text
-
Accepted Manuscript1.0
- Conference Paper:
- https://doi.org/10.1109/ITSC48978.2021.9564965
