Biological brains can effortlessly adapt to continuously changing stimulus environments, whereas conventional deep neural networks (DNNs) remain highly susceptible to domain shifts. Here, we demonstrate that static, hard-wired receptive fields, which spontaneously emerge in the early visual pathway, facilitate environment-agnostic object recognition in the brain. To test this mechanism, we introduced pre-developed Gabor filters in the early layers of DNNs, keeping them fixed during training. Despite the reduced learning flexibility, our networks exhibited robust continual learning capabilities under significant domain shifts, unlike conventional DNNs, which fail to generalize under similar conditions. Our network achieved generalized representations across domains in the latent space, while conventional DNNs only captured domain-specific variance. The static visual filters helped prevent local texture biases, leading to shape-based perception similar to that of primates. These findings highlight an intrinsic biological strategy that enables reliable continual learning in dynamic and unpredictable environments.

Event cameras provide microsecond latency, making them suitable for 6D object pose tracking in fast, dynamic scenes where conventional RGB and depth pipelines suffer from motion blur and large pixel displacements. We introduce EventTrack6D, an event-depth tracking framework that generalizes to novel objects without object-specific training by reconstructing both intensity and depth at arbitrary timestamps between depth frames. Conditioned on the most recent depth measurement, our dual reconstruction recovers dense photometric and geometric cues from sparse event streams. Our EventTrack6D operates at over 120 FPS and maintains temporal consistency under rapid motion. To support training and evaluation, we introduce a comprehensive benchmark suite: a large-scale synthetic dataset for training and two complementary evaluation sets, including real and simulated event datasets. Trained exclusively on synthetic data, EventTrack6D generalizes effectively to real-world scenarios without fine-tuning, maintaining accurate tracking across diverse objects and motion patterns. Our method and datasets establish event cameras as a viable solution for event-based novel object 6D pose tracking.

Novel view synthesis requires strong 3D geometric consistency and the ability to generate visually coherent images across diverse viewpoints. While recent camera-controlled video diffusion models show promising results, they often suffer from geometric distortions and limited camera controllability. To overcome these challenges, we introduce GeoNVS, a geometry-grounded novel-view synthesizer that enhances both geometric fidelity and camera controllability through explicit 3D geometric guidance. Our key innovation is the Gaussian Splat Feature Adapter (GS-Adapter), which lifts input-view diffusion features into 3D Gaussian representations, renders geometry-constrained novel-view features, and adaptively fuses them with diffusion features to correct geometrically inconsistent representations. Unlike prior methods that inject geometry at the input level, GS-Adapter operates in feature space, avoiding view-dependent color noise that degrades structural consistency. Its plug-and-play design enables zero-shot compatibility with diverse feed-forward geometry models without additional training, and can be adapted to other video diffusion backbones. Experiments across 9 scenes and 18 settings demonstrate state-of-the-art performance, achieving 11.3

Protein phase separation is a fundamental process in organizing membraneless organelles and is implicated in pathological conditions. Importantly, this process is dynamic and depends on conditions such as concentration, temperature, and solvent composition. However, current machine learning models infer phase separation propensity solely from amino acid sequences, failing to capture these context-dependent behaviors. Here we show that LLPSense, a machine learning framework that integrates protein language model embeddings with environmental parameters, achieves accurate, condition-aware predictions of phase separation. Multiple experimental validations confirm LLPSense's predictive power and utility. The model reveals complex, temperature-dependent reentrant behavior in SGTA, previously unrecognized as phase-separating. Moreover, LLPSense accurately predicts mutations in Parkinson's disease-associated α-synuclein that either enhance or suppress phase separation. Beyond predictive accuracy, model-guided mutagenesis enables the modulation of phase behavior. Collectively, LLPSense establishes a robust computational framework for interrogating phase landscapes, facilitating mechanistic disease studies and programmable condensate design.

We introduce Any6D, a model-free framework for 6D object pose estimation that requires only a single RGB-D anchor image to estimate both the 6D pose and size of unknown objects in novel scenes. Unlike existing methods that rely on textured 3D models or multiple viewpoints, Any6D leverages a joint object alignment process to enhance 2D-3D alignment and metric scale estimation for improved pose accuracy. Our approach integrates a render-and-compare strategy to generate and refine pose hypotheses, enabling robust performance in scenarios with occlusions, non-overlapping views, diverse lighting conditions, and large cross-environment variations. We evaluate our method on five challenging datasets: REAL275, Toyota-Light, HO3D, YCBINEOAT, and LM-O, demonstrating its effectiveness in significantly outperforming state-of-the-art methods for novel object pose estimation. Project page: https://taeyeop.com/any6d

We introduce Drag4D, an interactive framework that integrates object motion control within text-driven 3D scene generation. This framework enables users to define 3D trajectories for the 3D objects generated from a single image, seamlessly integrating them into a high-quality 3D background. Our Drag4D pipeline consists of three stages. First, we enhance text-to-3D background generation by applying 2D Gaussian Splatting with panoramic images and inpainted novel views, resulting in dense and visually complete 3D reconstructions. In the second stage, given a reference image of the target object, we introduce a 3D copy-and-paste approach: the target instance is extracted in a full 3D mesh using an off-the-shelf image-to-3D model and seamlessly composited into the generated 3D scene. The object mesh is then positioned within the 3D scene via our physics-aware object position learning, ensuring precise spatial alignment. Lastly, the spatially aligned object is temporally animated along a user-defined 3D trajectory. To mitigate motion hallucination and ensure view-consistent temporal alignment, we develop a part-augmented, motion-conditioned video diffusion model that processes multiview image pairs together with their projected 2D trajectories. We demonstrate the effectiveness of our unified architecture through evaluations at each stage and in the final results, showcasing the harmonized alignment of user-controlled object motion within a high-quality 3D background.

Conventional deep neural networks (DNNs) are highly susceptible to variations in input domains, unlike biological brains which effectively adapt to environmental changes. Here, we demonstrate that hard-wired Gabor filters, replicating the structure of receptive fields in the brains early visual pathway, facilitate environment-agnostic object recognition without overfitting. Our approach involved fixing the pre-designed Gabor filters in the early layers of DNNs, preventing any alterations during training. Despite the restricted learning flexibility of this model, our networks maintained robust performance even under significant domain shifts, in contrast to conventional DNNs that typically fail in similar conditions. We found that our model effectively clustered identical classes across diverse domains, while conventional DNNs tend to cluster images by domain in the latent space. We observed that the fixed Gabor filters enabled networks to encode global shape information rather than local texture features, thereby mitigating the risk of overfitting. ### Competing Interest Statement The authors have declared no competing interest.

This paper proposes an algorithm for synthesizing novel views under few-shot setup. The main concept is to develop a stable surface regularization technique called Annealing Signed Distance Function (ASDF), which anneals the surface in a coarse-to-fine manner to accelerate convergence speed. We observe that the Eikonal loss - which is a widely known geometric regularization - requires dense training signal to shape different level-sets of SDF, leading to low-fidelity results under few-shot training. In contrast, the proposed surface regularization successfully reconstructs scenes and produce high-fidelity geometry with stable training. Our method is further accelerated by utilizing grid representation and monocular geometric priors. Finally, the proposed approach is up to 45 times faster than existing few-shot novel view synthesis methods, and it produces comparable results in the ScanNet dataset and NeRF-Real dataset.

In this paper, we propose an end-to-end convolutional neural network for stereo matching with color and monochrome cameras, called CMSNet (Color and Monochrome Stereo Network). Both cameras have the same structure except for the presence of a Bayer filter, but have a fundamental trade-off. The Bayer filter allows capturing chrominance information of scenes, but limits a quantum efficiency of cameras, which causes severe image noise. It seems ideal if we can take advantage of both the cameras so that we obtain noise-free images with their corresponding disparity maps. However, image luminance recorded from a color camera is not consistent with that from a monochrome camera due to spatially-varying illumination and different spectral sensitivities of the cameras. This degrades the performance of stereo matching. To solve this problem, we design CMSNet for disparity estimation from noisy color and relatively clean monochrome images. CMSNet also infers a noise-free image with the estimated disparity map. We leverage a data augmentation to simulate realistic signal-dependent noise and various radiometric distortions between input stereo pairs to train CMSNet effectively. CMSNet is evaluated using various datasets and the performance of our disparity estimation and image enhancement consistently outperforms state-of-the-art methods.

Recently, Dual-Pixel (DP) sensors have been adopted in many imaging devices. However, despite their various advantages, DP sensors are used just for faster auto-focus and aesthetic image captures, and research on their usage for 3D facial understanding has been limited due to the lack of datasets and algorithmic designs that exploit parallax in DP images. It is also because the baseline of sub-aperture images is extremely narrow, and parallax exists in the defocus blur region. In this paper, we introduce a DP-oriented Depth/Normal estimation network that reconstructs the 3D facial geometry. In addition, to train the network, we collect DP facial data with more than 135K images for 101 persons captured with our multi-camera structured light systems. It contains ground-truth 3D facial models including depth map and surface normal in metric scale. Our dataset allows the proposed network to be generalized for 3D facial depth/normal estimation. The proposed network consists of two novel modules: Adaptive Sampling Module (ASM) and Adaptive Normal Module (ANM), which are specialized in handling the defocus blur in DP images. Finally, we demonstrate that the proposed method achieves state-of-the-art performances over recent DP-based depth/normal estimation methods.

To reconstruct a 3D scene from a set of calibrated views, traditional multi-view stereo techniques rely on two distinct stages: local depth maps computation and global depth maps fusion. Recent studies concentrate on deep neural architectures for depth estimation by using conventional depth fusion method or direct 3D reconstruction network by regressing Truncated Signed Distance Function (TSDF). In this paper, we advocate that replicating the traditional two stages framework with deep neural networks improves both the interpretability and the accuracy of the results. As mentioned, our network operates in two steps: 1) the local computation of the local depth maps with a deep MVS technique, and, 2) the depth maps and images' features fusion to build a single TSDF volume. In order to improve the matching performance between images acquired from very different viewpoints (e.g., large-baseline and rotations), we introduce a rotation-invariant 3D convolution kernel called PosedConv. The effectiveness of the proposed architecture is underlined via a large series of experiments conducted on the ScanNet dataset where our approach compares favorably against both traditional and deep learning techniques.