Diffusion models have achieved remarkable success in text-to-image synthesis, largely attributed to the use of classifier-free guidance (CFG), which enables high-quality, condition-aligned image generation. CFG combines the conditional score (e.g., text-conditioned) with the unconditional score to control the output. However, the unconditional score is in charge of estimating the transition between manifolds of adjacent timesteps from xtto xt−1, which may inadvertently interfere with the trajectory toward the specific condition. In this work, we introduce a novel approach that leverages a geometric perspective on the unconditional score to enhance CFG performance when conditional scores are available. Specifically, we propose a method that filters the singular vectors of both conditional and unconditional scores using singular value decomposition. This filtering process aligns the unconditional score with the conditional score, thereby refining the sampling trajectory to stay closer to the manifold. Our approach improves image quality with negligible additional computation. We provide deeper insights into the score function behavior in diffusion models and present a practical technique for achieving more accurate and contextually coherent image synthesis. project page

Trajectory prediction is a challenging problem that requires considering interactions among multiple actors and the surrounding environment. While data-driven approaches have been used to address this complex problem, they suffer from unreliable predictions under distribution shifts during test time. Accordingly, several online learning methods have been proposed using regression loss from the ground truth of observed data leveraging the auto-labeling nature of trajectory prediction task. We mainly tackle the following two issues. First, previous works underfit and overfit as they only optimize the last layer of motion decoder. To this end, we employ the masked autoencoder (MAE) for representation learning to encourage complex interaction modeling in shifted test distribution for updating deeper layers. Second, utilizing the sequential nature of driving data, we propose an actor-specific token memory that enables the test-time learning of actor-wise motion characteristics. Our proposed method has been validated across various challenging cross-dataset distribution shift scenarios including nuScenes, Lyft, Waymo, and Interaction. Our method surpasses the performance of existing state-of-the-art online learning methods in terms of both prediction accuracy and computational efficiency. The code is available at https://github.com/daeheepark/T4P.

Omni-directional images are becoming more prevalent for understanding the scene of all directions around a camera, as they provide a much wider field-of-view (FoV) compared to conventional images. In this work, we present a novel approach to represent omni-directional images and suggest how to apply CNNs on the proposed image representation. The proposed image representation method utilizes a spherical polyhedron to reduce distortion introduced inevitably when sampling pixels on a non-Euclidean spherical surface around the camera center. To apply convolution operation on our representation of images, we stack the neighboring pixels on top of each pixel and multiply with trainable parameters. This approach enables us to apply the same CNN architectures used in conventional euclidean 2D images on our proposed method in a straightforward manner. Compared to the previous work, we additionally compare different designs of kernels that can be applied to our proposed method. We also show that our method outperforms in monocular depth estimation task compared to other state-of-the-art representation methods of omni-directional images. In addition, we propose a novel method to fit bounding ellipses of arbitrary orientation using object detection networks and apply it to an omni-directional real-world human detection dataset.

Omni-directional cameras have many advantages over conventional cameras in that they have a much wider field-of-view (FOV). Accordingly, several approaches have been proposed recently to apply convolutional neural networks (CNNs) to omni-directional images for various visual tasks. However, most of them use image representations defined in the Euclidean space after transforming the omni-directional views originally formed in the non-Euclidean space. This transformation leads to shape distortion due to nonuniform spatial resolving power and the loss of continuity. These effects make existing convolution kernels experience difficulties in extracting meaningful information. This paper presents a novel method to resolve such problems of applying CNNs to omni-directional images. The proposed method utilizes a spherical polyhedron to represent omni-directional views. This method minimizes the variance of the spatial resolving power on the sphere surface, and includes new convolution and pooling methods for the proposed representation. The proposed method can also be adopted by any existing CNN-based methods. The feasibility of the proposed method is demonstrated through classification, detection, and semantic segmentation tasks with synthetic and real datasets.