skip to main content


Title: Image-to-Image Translation for Autonomous Driving from Coarsely-Aligned Image Pairs
A self-driving car must be able to reliably handle adverse weather conditions (e.g., snowy) to operate safely. In this paper, we investigate the idea of turning sensor inputs (i.e., images) captured in an adverse condition into a benign one (i.e., sunny), upon which the downstream tasks (e.g., semantic segmentation) can attain high accuracy. Prior work primarily formulates this as an unpaired image-to-image translation problem due to the lack of paired images captured under the exact same camera poses and semantic layouts. While perfectly- aligned images are not available, one can easily obtain coarsely- paired images. For instance, many people drive the same routes daily in both good and adverse weather; thus, images captured at close-by GPS locations can form a pair. Though data from repeated traversals are unlikely to capture the same foreground objects, we posit that they provide rich contextual information to supervise the image translation model. To this end, we propose a novel training objective leveraging coarsely- aligned image pairs. We show that our coarsely-aligned training scheme leads to a better image translation quality and improved downstream tasks, such as semantic segmentation, monocular depth estimation, and visual localization.  more » « less
Award ID(s):
2107161 1724282
NSF-PAR ID:
10466113
Author(s) / Creator(s):
; ; ; ; ;
Date Published:
Journal Name:
Proceedings IEEE International Conference on Robotics and Automation
ISSN:
1050-4729
Page Range / eLocation ID:
7756-7762
Format(s):
Medium: X
Sponsoring Org:
National Science Foundation
More Like this
  1. Image-to-image translation is a class of vision and graphics problems where the goal is to learn the mapping between an input image and an output image using a training set of aligned image pairs. However, for many tasks, paired training data will not be available. We present an approach for learning to translate an image from a source domain $X$ to a target domain $Y$ in the absence of paired examples. Our goal is to learn a mapping $G: X \rightarrow Y$ such that the distribution of images from $G(X)$ is indistinguishable from the distribution $Y$ using an adversarial loss. Because this mapping is highly under-constrained, we couple it with an inverse mapping $F: Y \rightarrow X$ and introduce a {\em cycle consistency loss} to push $F(G(X)) \approx X$ (and vice versa). Qualitative results are presented on several tasks where paired training data does not exist, including collection style transfer, object transfiguration, season transfer, photo enhancement, etc. Quantitative comparisons against several prior methods demonstrate the superiority of our approach. 
    more » « less
  2. We propose a new approach for high resolution semantic image synthesis. It consists of one base image generator and multiple class-specific generators. The base generator generates high quality images based on a segmentation map. To further improve the quality of different objects, we create a bank of Generative Adversarial Networks (GANs) by separately training class-specific models. This has several benefits including – dedicated weights for each class; centrally aligned data for each model; additional training data from other sources, potential of higher resolution and quality; and easy manipulation of a specific object in the scene. Experiments show that our approach can generate high quality images in high resolution while having flexibility of object-level control by using class-specific generators. Project page: https://yuheng-li.github.io/CollageGAN/ 
    more » « less
  3. An unsupervised image-to-image translation (UI2I) task deals with learning a mapping between two domains without paired images. While existing UI2I methods usually require numerous unpaired images from different domains for training, there are many scenarios where training data is quite limited. In this paper, we argue that even if each domain contains a single image, UI2I can still be achieved. To this end, we propose TuiGAN, a generative model that is trained on only two unpaired images and amounts to one-shot unsupervised learning. With TuiGAN, an image is translated in a coarse-to-fine manner where the generated image is gradually refined from global structures to local details. We conduct extensive experiments to verify that our versatile method can outperform strong baselines on a wide variety of UI2I tasks. Moreover, TuiGAN is capable of achieving comparable performance with the state-of-the-art UI2I models trained with sufficient data. 
    more » « less
  4. Transformers have shown great success in medical image segmentation. However, transformers may exhibit a limited generalization ability due to the underlying single-scale selfattention (SA) mechanism. In this paper, we address this issue by introducing a Multiscale hiERarchical vIsion Transformer (MERIT) backbone network, which improves the generalizability of the model by computing SA at multiple scales. We also incorporate an attention-based decoder, namely Cascaded Attention Decoding (CASCADE), for further refinement of the multi-stage features generated by MERIT. Finally, we introduce an effective multi-stage feature mixing loss aggregation (MUTATION) method for better model training via implicit ensembling. Our experiments on two widely used medical image segmentation benchmarks (i.e., Synapse Multi-organ and ACDC) demonstrate the superior performance of MERIT over state-of-the-art methods. Our MERIT architecture and MUTATION loss aggregation can be used with other downstream medical image and semantic segmentation tasks. 
    more » « less
  5. Sketch-to-image is an important task to reduce the burden of creating a color image from scratch. Unlike previous sketch-to-image models, where the image is synthesized in an end-to-end manner, leading to an unnaturalistic image, we propose a method by decomposing the problem into subproblems to generate a more naturalistic and reasonable image. It first generates an intermediate output which is a semantic mask map from the input sketch through instance and semantic segmentation in two levels, background segmentation and foreground segmentation. Background segmentation is formed based on the context of the foreground objects. Then, the foreground segmentations are sequentially added to the created background segmentation. Finally, the generated mask map is fed into an image-to-image translation model to generate an image. Our proposed method works with 92 distinct classes. Compared to state-of-the-art sketch-to-image models, our proposed method outperforms the previous methods and generates better images. 
    more » « less