UMD Computer Vision Seminar

Abstract: Atmospheric turbulence presents a significant challenge in long-range imaging. Current restoration algorithms often struggle with temporal inconsistency, as well as limited generalization ability across varying turbulence levels and scene content different than the training data. To tackle these issues, we introduce a self-supervised method, Consistent Video Restoration through Turbulence (ConVRT) a test-time optimization method featuring a neural video representation designed to enhance temporal consistency in restoration. A key innovation of ConVRT is the integration of a pretrained vision-language model (CLIP) for semantic-oriented supervision, which steers the restoration towards sharp, photorealistic images in the CLIP latent space. We further develop a principled selection strategy of text prompts, based on their statistical correlation with a perceptual metric. ConVRT's test-time optimization allows it to adapt to a wide range of real-world turbulence conditions, effectively leveraging the insights gained from pre-trained models on simulated data. ConVRT offers a comprehensive and effective solution for mitigating real-world turbulence in dynamic videos.

Abstract: We provide the first perceptual quantification of user's sensitivity to radial optic flow artifacts and demonstrate a promising approach for masking this optic flow artifact via blink suppression. Near-eye HMDs allow users to feel immersed in virtual environments by providing visual cues, like motion parallax and stereoscopy, that mimic how we view the physical world. However, these systems exhibit a variety of perceptual artifacts that can limit their usability and the user's sense of presence in VR. One well-known artifact is the vergence-accommodation conflict (VAC). Varifocal displays can mitigate VAC, but bring with them other artifacts such as a change in virtual image size (radial optic flow) when the focal plane changes. We conducted a set of psychophysical studies to measure users' ability to perceive this radial flow artifact before, during, and after self-initiated blinks. Our results showed that visual sensitivity was reduced by a factor of 10 at the start and for ~70 ms after a blink was detected. Pre- and post-blink sensitivity was, on average, ~0.15% image size change during normal viewing and increased to ~1.5-2.0% during blinks. Our results imply that a rapid (under 70 ms) radial optic flow distortion can go unnoticed during a blink. Furthermore, our results provide empirical data that can be used to inform engineering requirements for both hardware design and software-based graphical correction algorithms for future varifocal near-eye displays.

Abstract: Scene-aware Complementary Item Retrieval (CIR) is a challenging task which requires to generate a set of compatible items across domains. Due to the subjectivity, it is difficult to set up a rigorous standard for both data collection and learning objectives. To address this challenging task, we propose a visual compatibility concept, composed of similarity (resembling in color, geometry, texture, and etc.) and complementarity (different items like table vs chair completing a group). Based on this notion, we propose a compatibility learning framework, a category-aware Flexible Bidirectional Transformer (FBT), for visual "scene-based set compatibility reasoning" with the cross-domain visual similarity input and auto-regressive complementary item generation. We introduce a "Flexible Bidirectional Transformer (FBT)," consisting of an encoder with flexible masking, a category prediction arm, and an auto-regressive visual embedding prediction arm. And the inputs for FBT are cross-domain visual similarity invariant embeddings, making this framework quite generalizable. Furthermore, our proposed FBT model learns the inter-object compatibility from a large set of scene images in a self-supervised way. Compared with the SOTA methods, this approach achieves up to 5.3% and 9.6% in FITB score and 22.3% and 31.8% SFID improvement on fashion and furniture, respectively.

Abstract: We introduce a simple yet effective approach for separating transmitted and reflected light. Our key insight is that the powerful novel view synthesis capabilities provided by modern inverse rendering methods (e.g.,~Gaussian splatting) allow one to perform flash/no-flash reflection separation using unpaired measurements---this relaxation dramatically simplifies image acquisition over conventional paired flash/no-flash reflection separation methods. Through extensive real-world experiments, we demonstrate our method, Flash-Splat, accurately reconstructs both transmitted and reflected scenes in 3D. Our method outperforms existing 3D reflection separation methods, which do not leverage illumination control, by a large margin.

Abstract: Novel view synthesis from an in-the-wild video is difficult due to challenges like scene dynamics and lack of parallax. While existing methods have shown promising results with implicit neural radiance fields, they are slow to train and render. This paper revisits explicit video representations to synthesize high-quality novel views from a monocular video efficiently. We treat static and dynamic video content separately. Specifically, we build a global static scene model using an extended plane-based scene representation to synthesize temporally coherent novel video. Our plane-based scene representation is augmented with spherical harmonics and displacement maps to capture view-dependent effects and model non-planar complex surface geometry. We opt to represent the dynamic content as per-frame point clouds for efficiency. While such representations are inconsistency-prone, minor temporal inconsistencies are perceptually masked due to motion. We develop a method to quickly estimate such a hybrid video representation and render novel views in real time. Our experiments show that our method can render high-quality novel views from an in-the-wild video with comparable quality to state-of-the-art methods while being 100x faster in training and enabling real-time rendering. Arxiv.

Abstract: Vision Transformers (ViTs) have gained significant popularity in recent years and have proliferated into many applications. However, it is not well explored how varied their behavior is under different learning paradigms. We compare ViTs trained through different methods of supervision, and show that they learn a diverse range of behaviors in terms of their attention, representations, and downstream performance. We also discover ViT behaviors that are consistent across supervision, including the emergence of Offset Local Attention Heads. These are self-attention heads that attend to a token adjacent to the current token with a fixed directional offset, a phenomenon that to the best of our knowledge has not been highlighted in any prior work. Our analysis shows that ViTs are highly flexible and learn to process local and global information in different orders depending on their training method. We find that contrastive self-supervised methods learn features that are competitive with explicitly supervised features, and they can even be superior for part-level tasks. We also find that the representations of reconstruction-based models show non-trivial similarity to contrastive self-supervised models. Finally, we show how the "best" layer for a given task varies by both supervision method and task, further demonstrating the differing order of information processing in ViTs. Arxiv.

Abstract: Creating 3D models from casual camera captures enables various creative and cognitive applications for AR/VR, Computational Photography, Robotics, and Medical Imaging. Nonetheless, the underlying 'engines' propelling these 3D models comprising the hardware systems for capture and machine learning models are often expensive to build, thereby constraining widespread accessibility. In this talk, I will discuss my group's effort to develop efficient and accessible 3D models. The first part of the talk will focus on developing lightweight and personalized 3D generative models for facial reconstruction and relighting. The last part of the talk will focus on developing efficient 3D reconstruction algorithms that combine both camera and lighting variations for objects, scenes, and endoscopy images.

Abstract: We propose a generative model that, given a coarsely edited image, synthesizes a photorealistic output that follows the prescribed layout. Our method transfers fine details from the original image and preserve the identity of its parts. Yet, it adapts it to the lighting and context defined by the new layout. Our key insight is that videos are a powerful source of supervision for this task: objects and camera motions provide many observations of how the world changes with viewpoint, lighting, and physical interactions. We construct an image dataset in which each sample is a pair of source and target frames extracted from the same video at randomly chosen time intervals. We warp the source frame toward the target using two motion models that mimic the expected test-time user edits. We supervise our model to translate the warped image into the ground truth, starting from a pretrained diffusion model. Our model design explicitly enables fine detail transfer from the source frame to the generated image, while closely following the user-specified layout. We show that by using simple segmentations and coarse 2D manipulations, we can synthesize a photorealistic edit faithful to the user's input while addressing second-order effects like harmonizing the lighting and physical interactions between edited objects.

Abstract: Digitization of reality is at the cusp of widespread adoption and 3D displays are at the forefront of enabling such extended reality systems. For an immersive experience, a 3D display must faithfully reproduce the visual cues pertaining to vergence, accommodation, occlusion, and motion parallax. I will talk about recent work on computational display designs that enable such features in near-eye displays.

Abstract: Underwater perception and 3D surface reconstruction are challenging problems with broad applications in construction, security, marine archaeology, and environmental monitoring. Treacherous operating conditions, fragile surroundings, and limited navigation control often dictate that submersibles restrict their range of motion and, thus, the baseline over which they can capture measurements. In the context of 3D scene reconstruction, it is well-known that smaller baselines make reconstruction more challenging. Our work develops a physics-based multimodal acoustic-optical neural surface reconstruction framework capable of effectively integrating high-resolution RGB measurements with low-resolution depth-resolved imaging sonar measurements. By fusing these complementary modalities, our framework can reconstruct accurate high-resolution 3D surfaces from measurements captured over heavily-restricted baselines. Through extensive simulations and in-lab experiments, we demonstrate that our method dramatically outperforms recent RGB-only and sonar-only inverse-differentiable-rendering-based surface reconstruction methods.