Yung-Yu Chuang

Yu-Hsi Chen, Chien-Yao Wang, Cheng-Yun Yang et al.

We propose a post-processor, called NeighborTrack, that leverages neighbor information of the tracking target to validate and improve single-object tracking (SOT) results. It requires no additional data or retraining. Instead, it uses the confidence score predicted by the backbone SOT network to automatically derive neighbor information and then uses this information to improve the tracking results. When tracking an occluded target, its appearance features are untrustworthy. However, a general siamese network often cannot tell whether the tracked object is occluded by reading the confidence score alone, because it could be misled by neighbors with high confidence scores. Our proposed NeighborTrack takes advantage of unoccluded neighbors' information to reconfirm the tracking target and reduces false tracking when the target is occluded. It not only reduces the impact caused by occlusion, but also fixes tracking problems caused by object appearance changes. NeighborTrack is agnostic to SOT networks and post-processing methods. For the VOT challenge dataset commonly used in short-term object tracking, we improve three famous SOT networks, Ocean, TransT, and OSTrack, by an average of ${1.92\%}$ EAO and ${2.11\%}$ robustness. For the mid- and long-term tracking experiments based on OSTrack, we achieve state-of-the-art ${72.25\%}$ AUC on LaSOT and ${75.7\%}$ AO on GOT-10K. Code duplication can be found in https://github.com/franktpmvu/NeighborTrack.

Ziyuan Huang, Zhengping Zhou, Yung-Yu Chuang et al. · stanford

We present a new method for generating controllable, dynamically responsive, and photorealistic human animations. Given an image of a person, our system allows the user to generate Physically plausible Upper Body Animation (PUBA) using interaction in the image space, such as dragging their hand to various locations. We formulate a reinforcement learning problem to train a dynamic model that predicts the person's next 2D state (i.e., keypoints on the image) conditioned on a 3D action (i.e., joint torque), and a policy that outputs optimal actions to control the person to achieve desired goals. The dynamic model leverages the expressiveness of 3D simulation and the visual realism of 2D videos. PUBA generates 2D keypoint sequences that achieve task goals while being responsive to forceful perturbation. The sequences of keypoints are then translated by a pose-to-image generator to produce the final photorealistic video.

Yu-Lun Liu, Chen Gao, Andreas Meuleman et al.

Dynamic radiance field reconstruction methods aim to model the time-varying structure and appearance of a dynamic scene. Existing methods, however, assume that accurate camera poses can be reliably estimated by Structure from Motion (SfM) algorithms. These methods, thus, are unreliable as SfM algorithms often fail or produce erroneous poses on challenging videos with highly dynamic objects, poorly textured surfaces, and rotating camera motion. We address this robustness issue by jointly estimating the static and dynamic radiance fields along with the camera parameters (poses and focal length). We demonstrate the robustness of our approach via extensive quantitative and qualitative experiments. Our results show favorable performance over the state-of-the-art dynamic view synthesis methods.

Zhong-Min Tsai, Yu-Ju Tsai, Chien-Yao Wang et al.

The paper presents a new method, SearchTrack, for multiple object tracking and segmentation (MOTS). To address the association problem between detected objects, SearchTrack proposes object-customized search and motion-aware features. By maintaining a Kalman filter for each object, we encode the predicted motion into the motion-aware feature, which includes both motion and appearance cues. For each object, a customized fully convolutional search engine is created by SearchTrack by learning a set of weights for dynamic convolutions specific to the object. Experiments demonstrate that our SearchTrack method outperforms competitive methods on both MOTS and MOT tasks, particularly in terms of association accuracy. Our method achieves 71.5 HOTA (car) and 57.6 HOTA (pedestrian) on the KITTI MOTS and 53.4 HOTA on MOT17. In terms of association accuracy, our method achieves state-of-the-art performance among 2D online methods on the KITTI MOTS. Our code is available at https://github.com/qa276390/SearchTrack.

Zheng-Hui Huang, Zhixiang Wang, Yu-Lun Liu et al.

Single-image reflection separation is highly challenging under extreme conditions like glare or weak reflections. Existing methods often struggle to recover both layers in glare or weak-reflection scenarios because of insufficient information. This paper presents a diffusion model explicitly fine-tuned for this task, leveraging generative diffusion priors for robust separation. Our method simultaneously generates transmission and reflection layers through a unified diffusion model, incorporating a novel cross-layer self-attention mechanism for better feature disentanglement. We further introduce a disjoint sampling strategy to iteratively reduce interference between the layers during diffusion and a latent optimization step with a learned composition function for improved results in complex real-world scenarios. Extensive experiments demonstrate that our approach surpasses state-of-the-art methods on multiple real-world benchmarks. Project page: https://brian90709.github.io/diff-reflection-separation/

Zhixiang Wang, Baiang Li, Jian Wang et al.

This paper introduces an innovative approach for image matting that redefines the traditional regression-based task as a generative modeling challenge. Our method harnesses the capabilities of latent diffusion models, enriched with extensive pre-trained knowledge, to regularize the matting process. We present novel architectural innovations that empower our model to produce mattes with superior resolution and detail. The proposed method is versatile and can perform both guidance-free and guidance-based image matting, accommodating a variety of additional cues. Our comprehensive evaluation across three benchmark datasets demonstrates the superior performance of our approach, both quantitatively and qualitatively. The results not only reflect our method's robust effectiveness but also highlight its ability to generate visually compelling mattes that approach photorealistic quality. The project page for this paper is available at https://lightchaserx.github.io/matting-by-generation/

Cheng-Kun Yang, Min-Hung Chen, Yung-Yu Chuang et al.

We present a Multimodal Interlaced Transformer (MIT) that jointly considers 2D and 3D data for weakly supervised point cloud segmentation. Research studies have shown that 2D and 3D features are complementary for point cloud segmentation. However, existing methods require extra 2D annotations to achieve 2D-3D information fusion. Considering the high annotation cost of point clouds, effective 2D and 3D feature fusion based on weakly supervised learning is in great demand. To this end, we propose a transformer model with two encoders and one decoder for weakly supervised point cloud segmentation using only scene-level class tags. Specifically, the two encoders compute the self-attended features for 3D point clouds and 2D multi-view images, respectively. The decoder implements interlaced 2D-3D cross-attention and carries out implicit 2D and 3D feature fusion. We alternately switch the roles of queries and key-value pairs in the decoder layers. It turns out that the 2D and 3D features are iteratively enriched by each other. Experiments show that it performs favorably against existing weakly supervised point cloud segmentation methods by a large margin on the S3DIS and ScanNet benchmarks. The project page will be available at https://jimmy15923.github.io/mit_web/.

Ping-Mao Huang, I-Tien Chao, Ping-Chia Huang et al.

Real-time semantic segmentation presents the dual challenge of designing efficient architectures that capture large receptive fields for semantic understanding while also refining detailed contours. Vision transformers model long-range dependencies effectively but incur high computational cost. To address these challenges, we introduce the Large Kernel Attention (LKA) mechanism. Our proposed Bilateral Efficient Visual Attention Network (BEVANet) expands the receptive field to capture contextual information and extracts visual and structural features using Sparse Decomposed Large Separable Kernel Attentions (SDLSKA). The Comprehensive Kernel Selection (CKS) mechanism dynamically adapts the receptive field to further enhance performance. Furthermore, the Deep Large Kernel Pyramid Pooling Module (DLKPPM) enriches contextual features by synergistically combining dilated convolutions and large kernel attention. The bilateral architecture facilitates frequent branch communication, and the Boundary Guided Adaptive Fusion (BGAF) module enhances boundary delineation by integrating spatial and semantic features under boundary guidance. BEVANet achieves real-time segmentation at 33 FPS, yielding 79.3% mIoU without pretraining and 81.0% mIoU on Cityscapes after ImageNet pretraining, demonstrating state-of-the-art performance. The code and model is available at https://github.com/maomao0819/BEVANet.

Ji-Jia Wu, Andy Chia-Hao Chang, Chieh-Yu Chuang et al.

This paper addresses text-supervised semantic segmentation, aiming to learn a model capable of segmenting arbitrary visual concepts within images by using only image-text pairs without dense annotations. Existing methods have demonstrated that contrastive learning on image-text pairs effectively aligns visual segments with the meanings of texts. We notice that there is a discrepancy between text alignment and semantic segmentation: A text often consists of multiple semantic concepts, whereas semantic segmentation strives to create semantically homogeneous segments. To address this issue, we propose a novel framework, Image-Text Co-Decomposition (CoDe), where the paired image and text are jointly decomposed into a set of image regions and a set of word segments, respectively, and contrastive learning is developed to enforce region-word alignment. To work with a vision-language model, we present a prompt learning mechanism that derives an extra representation to highlight an image segment or a word segment of interest, with which more effective features can be extracted from that segment. Comprehensive experimental results demonstrate that our method performs favorably against existing text-supervised semantic segmentation methods on six benchmark datasets.

Zheng-Hui Huang, Zhixiang Wang, Jiaming Tan et al.

Scaling generative inverse and forward rendering to real-world scenarios is bottlenecked by the limited realism and temporal coherence of existing synthetic datasets. To bridge this persistent domain gap, we introduce a large-scale, dynamic dataset curated from visually complex AAA games. Using a novel dual-screen stitched capture method, we extracted 4M continuous frames (720p/30 FPS) of synchronized RGB and five G-buffer channels across diverse scenes, visual effects, and environments, including adverse weather and motion-blur variants. This dataset uniquely advances bidirectional rendering: enabling robust in-the-wild geometry and material decomposition, and facilitating high-fidelity G-buffer-guided video generation. Furthermore, to evaluate the real-world performance of inverse rendering without ground truth, we propose a novel VLM-based assessment protocol measuring semantic, spatial, and temporal consistency. Experiments demonstrate that inverse renderers fine-tuned on our data achieve superior cross-dataset generalization and controllable generation, while our VLM evaluation strongly correlates with human judgment. Combined with our toolkit, our forward renderer enables users to edit styles of AAA games from G-buffers using text prompts.

Yi-Ting Pan, Chai-Rong Lee, Shu-Ho Fan et al.

The entertainment industry relies on 3D visual content to create immersive experiences, but traditional methods for creating textured 3D models can be time-consuming and subjective. Generative networks such as StyleGAN have advanced image synthesis, but generating 3D objects with high-fidelity textures is still not well explored, and existing methods have limitations. We propose the Semantic-guided Conditional Texture Generator (CTGAN), producing high-quality textures for 3D shapes that are consistent with the viewing angle while respecting shape semantics. CTGAN utilizes the disentangled nature of StyleGAN to finely manipulate the input latent codes, enabling explicit control over both the style and structure of the generated textures. A coarse-to-fine encoder architecture is introduced to enhance control over the structure of the resulting textures via input segmentation. Experimental results show that CTGAN outperforms existing methods on multiple quality metrics and achieves state-of-the-art performance on texture generation in both conditional and unconditional settings.

Yu-Lun Liu, Wei-Sheng Lai, Ming-Hsuan Yang et al.

Existing video stabilization methods often generate visible distortion or require aggressive cropping of frame boundaries, resulting in smaller field of views. In this work, we present a frame synthesis algorithm to achieve full-frame video stabilization. We first estimate dense warp fields from neighboring frames and then synthesize the stabilized frame by fusing the warped contents. Our core technical novelty lies in the learning-based hybrid-space fusion that alleviates artifacts caused by optical flow inaccuracy and fast-moving objects. We validate the effectiveness of our method on the NUS, selfie, and DeepStab video datasets. Extensive experiment results demonstrate the merits of our approach over prior video stabilization methods.

Yu-Lun Liu, Wei-Sheng Lai, Ming-Hsuan Yang et al.

We present a learning-based approach for removing unwanted obstructions, such as window reflections, fence occlusions, or adherent raindrops, from a short sequence of images captured by a moving camera. Our method leverages motion differences between the background and obstructing elements to recover both layers. Specifically, we alternate between estimating dense optical flow fields of the two layers and reconstructing each layer from the flow-warped images via a deep convolutional neural network. This learning-based layer reconstruction module facilitates accommodating potential errors in the flow estimation and brittle assumptions, such as brightness consistency. We show that the proposed approach learned from synthetically generated data performs well to real images. Experimental results on numerous challenging scenarios of reflection and fence removal demonstrate the effectiveness of the proposed method.

Hsin-Yu Chang, Zhixiang Wang, Yung-Yu Chuang

Style transfer generates an image whose content comes from one image and style from the other. Image-to-image translation approaches with disentangled representations have been shown effective for style transfer between two image categories. However, previous methods often assume a shared domain-invariant content space, which could compromise the content representation power. For addressing this issue, this paper leverages domain-specific mappings for remapping latent features in the shared content space to domain-specific content spaces. This way, images can be encoded more properly for style transfer. Experiments show that the proposed method outperforms previous style transfer methods, particularly on challenging scenarios that would require semantic correspondences between images. Code and results are available at https://acht7111020.github.io/DSMAP-demo/.

Sheng-Yeh Chen, Yung-Yu Chuang

Modern cameras have limited dynamic ranges and often produce images with saturated or dark regions using a single exposure. Although the problem could be addressed by taking multiple images with different exposures, exposure fusion methods need to deal with ghosting artifacts and detail loss caused by camera motion or moving objects. This paper proposes a deep network for exposure fusion. For reducing the potential ghosting problem, our network only takes two images, an underexposed image and an overexposed one. Our network integrates together homography estimation for compensating camera motion, attention mechanism for correcting remaining misalignment and moving pixels, and adversarial learning for alleviating other remaining artifacts. Experiments on real-world photos taken using handheld mobile phones show that the proposed method can generate high-quality images with faithful detail and vivid color rendition in both dark and bright areas.

Yu-Lun Liu, Wei-Sheng Lai, Ming-Hsuan Yang et al.

We present a learning-based approach for removing unwanted obstructions, such as window reflections, fence occlusions or raindrops, from a short sequence of images captured by a moving camera. Our method leverages the motion differences between the background and the obstructing elements to recover both layers. Specifically, we alternate between estimating dense optical flow fields of the two layers and reconstructing each layer from the flow-warped images via a deep convolutional neural network. The learning-based layer reconstruction allows us to accommodate potential errors in the flow estimation and brittle assumptions such as brightness consistency. We show that training on synthetically generated data transfers well to real images. Our results on numerous challenging scenarios of reflection and fence removal demonstrate the effectiveness of the proposed method.

Yu-Lun Liu, Wei-Sheng Lai, Yu-Sheng Chen et al.

Recovering a high dynamic range (HDR) image from a single low dynamic range (LDR) input image is challenging due to missing details in under-/over-exposed regions caused by quantization and saturation of camera sensors. In contrast to existing learning-based methods, our core idea is to incorporate the domain knowledge of the LDR image formation pipeline into our model. We model the HDRto-LDR image formation pipeline as the (1) dynamic range clipping, (2) non-linear mapping from a camera response function, and (3) quantization. We then propose to learn three specialized CNNs to reverse these steps. By decomposing the problem into specific sub-tasks, we impose effective physical constraints to facilitate the training of individual sub-networks. Finally, we jointly fine-tune the entire model end-to-end to reduce error accumulation. With extensive quantitative and qualitative experiments on diverse image datasets, we demonstrate that the proposed method performs favorably against state-of-the-art single-image HDR reconstruction algorithms.

Zelong Zeng, Zhixiang Wang, Zheng Wang et al.

Most person re-identification (ReID) approaches assume that person images are captured under relatively similar illumination conditions. In reality, long-term person retrieval is common, and person images are often captured under different illumination conditions at different times across a day. In this situation, the performances of existing ReID models often degrade dramatically. This paper addresses the ReID problem with illumination variations and names it as {\em Illumination-Adaptive Person Re-identification (IA-ReID)}. We propose an Illumination-Identity Disentanglement (IID) network to dispel different scales of illuminations away while preserving individuals' identity information. To demonstrate the illumination issue and to evaluate our model, we construct two large-scale simulated datasets with a wide range of illumination variations. Experimental results on the simulated datasets and real-world images demonstrate the effectiveness of the proposed framework.

Nai-Sheng Syu, Yu-Sheng Chen, Yung-Yu Chuang

This paper presents a comprehensive study of applying the convolutional neural network (CNN) to solving the demosaicing problem. The paper presents two CNN models that learn end-to-end mappings between the mosaic samples and the original image patches with full information. In the case the Bayer color filter array (CFA) is used, an evaluation with ten competitive methods on popular benchmarks confirms that the data-driven, automatically learned features by the CNN models are very effective. Experiments show that the proposed CNN models can perform equally well in both the sRGB space and the linear space. It is also demonstrated that the CNN model can perform joint denoising and demosaicing. The CNN model is very flexible and can be easily adopted for demosaicing with any CFA design. We train CNN models for demosaicing with three different CFAs and obtain better results than existing methods. With the great flexibility to be coupled with any CFA, we present the first data-driven joint optimization of the CFA design and the demosaicing method using CNN. Experiments show that the combination of the automatically discovered CFA pattern and the automatically devised demosaicing method significantly outperforms the current best demosaicing results. Visual comparisons confirm that the proposed methods reduce more visual artifacts than existing methods. Finally, we show that the CNN model is also effective for the more general demosaicing problem with spatially varying exposure and color and can be used for taking images of higher dynamic ranges with a single shot. The proposed models and the thorough experiments together demonstrate that CNN is an effective and versatile tool for solving the demosaicing problem.