Suya You

Cody Simons, Dripta S. Raychaudhuri, Sk Miraj Ahmed et al.

Scene understanding using multi-modal data is necessary in many applications, e.g., autonomous navigation. To achieve this in a variety of situations, existing models must be able to adapt to shifting data distributions without arduous data annotation. Current approaches assume that the source data is available during adaptation and that the source consists of paired multi-modal data. Both these assumptions may be problematic for many applications. Source data may not be available due to privacy, security, or economic concerns. Assuming the existence of paired multi-modal data for training also entails significant data collection costs and fails to take advantage of widely available freely distributed pre-trained uni-modal models. In this work, we relax both of these assumptions by addressing the problem of adapting a set of models trained independently on uni-modal data to a target domain consisting of unlabeled multi-modal data, without having access to the original source dataset. Our proposed approach solves this problem through a switching framework which automatically chooses between two complementary methods of cross-modal pseudo-label fusion -- agreement filtering and entropy weighting -- based on the estimated domain gap. We demonstrate our work on the semantic segmentation problem. Experiments across seven challenging adaptation scenarios verify the efficacy of our approach, achieving results comparable to, and in some cases outperforming, methods which assume access to source data. Our method achieves an improvement in mIoU of up to 12% over competing baselines. Our code is publicly available at https://github.com/csimo005/SUMMIT.

Yunhao Ba, Howard Zhang, Ethan Yang et al.

We propose a large-scale dataset of real-world rainy and clean image pairs and a method to remove degradations, induced by rain streaks and rain accumulation, from the image. As there exists no real-world dataset for deraining, current state-of-the-art methods rely on synthetic data and thus are limited by the sim2real domain gap; moreover, rigorous evaluation remains a challenge due to the absence of a real paired dataset. We fill this gap by collecting a real paired deraining dataset through meticulous control of non-rain variations. Our dataset enables paired training and quantitative evaluation for diverse real-world rain phenomena (e.g. rain streaks and rain accumulation). To learn a representation robust to rain phenomena, we propose a deep neural network that reconstructs the underlying scene by minimizing a rain-robust loss between rainy and clean images. Extensive experiments demonstrate that our model outperforms the state-of-the-art deraining methods on real rainy images under various conditions. Project website: https://visual.ee.ucla.edu/gt_rain.htm/.

Yang You, Mikaela Angelina Uy, Jiaqi Han et al.

Reverse engineering 3D computer-aided design (CAD) models from images is an important task for many downstream applications including interactive editing, manufacturing, architecture, robotics, etc. The difficulty of the task lies in vast representational disparities between the CAD output and the image input. CAD models are precise, programmatic constructs that involve sequential operations combining discrete command structure with continuous attributes, making it challenging to learn and optimize in an end-to-end fashion. Concurrently, input images introduce inherent challenges such as photometric variability and sensor noise, complicating the reverse engineering process. In this work, we introduce a novel approach that conditionally factorizes the task into two sub-problems. First, we leverage vision-language foundation models (VLMs), a finetuned Llama3.2, to predict the global discrete base structure with semantic information. Second, we propose TrAssembler that, conditioned on the discrete structure with semantics, predicts the continuous attribute values. To support the training of our TrAssembler, we further constructed an annotated CAD dataset of common objects from ShapeNet. Putting all together, our approach and data demonstrate significant first steps towards CAD-ifying images in the wild. Code and data can be found in https://github.com/qq456cvb/Img2CAD.

Zhen Wang, Shijie Zhou, Jeong Joon Park et al.

This work introduces alternating latent topologies (ALTO) for high-fidelity reconstruction of implicit 3D surfaces from noisy point clouds. Previous work identifies that the spatial arrangement of latent encodings is important to recover detail. One school of thought is to encode a latent vector for each point (point latents). Another school of thought is to project point latents into a grid (grid latents) which could be a voxel grid or triplane grid. Each school of thought has tradeoffs. Grid latents are coarse and lose high-frequency detail. In contrast, point latents preserve detail. However, point latents are more difficult to decode into a surface, and quality and runtime suffer. In this paper, we propose ALTO to sequentially alternate between geometric representations, before converging to an easy-to-decode latent. We find that this preserves spatial expressiveness and makes decoding lightweight. We validate ALTO on implicit 3D recovery and observe not only a performance improvement over the state-of-the-art, but a runtime improvement of 3-10$\times$. Project website at https://visual.ee.ucla.edu/alto.htm/.

Zhikang Zhang, Zifan Yu, Suya You et al. · amazon-science

Motivated by the increasing application of low-resolution LiDAR recently, we target the problem of low-resolution LiDAR-camera calibration in this work. The main challenges are two-fold: sparsity and noise in point clouds. To address the problem, we propose to apply depth interpolation to increase the point density and supervised contrastive learning to learn noise-resistant features. The experiments on RELLIS-3D demonstrate that our approach achieves an average mean absolute rotation/translation errors of 0.15cm/0.33\textdegree on 32-channel LiDAR point cloud data, which significantly outperforms all reference methods.

Vishal Asnani, Abhinav Kumar, Suya You et al.

Previous research in $2D$ object detection focuses on various tasks, including detecting objects in generic and camouflaged images. These works are regarded as passive works for object detection as they take the input image as is. However, convergence to global minima is not guaranteed to be optimal in neural networks; therefore, we argue that the trained weights in the object detector are not optimal. To rectify this problem, we propose a wrapper based on proactive schemes, PrObeD, which enhances the performance of these object detectors by learning a signal. PrObeD consists of an encoder-decoder architecture, where the encoder network generates an image-dependent signal termed templates to encrypt the input images, and the decoder recovers this template from the encrypted images. We propose that learning the optimum template results in an object detector with an improved detection performance. The template acts as a mask to the input images to highlight semantics useful for the object detector. Finetuning the object detector with these encrypted images enhances the detection performance for both generic and camouflaged. Our experiments on MS-COCO, CAMO, COD$10$K, and NC$4$K datasets show improvement over different detectors after applying PrObeD. Our models/codes are available at https://github.com/vishal3477/Proactive-Object-Detection.

Zhiruo Zhou, Suya You, C. -C. Jay Kuo

Supervised trackers trained on labeled data dominate the single object tracking field for superior tracking accuracy. The labeling cost and the huge computational complexity hinder their applications on edge devices. Unsupervised learning methods have also been investigated to reduce the labeling cost but their complexity remains high. Aiming at lightweight high-performance tracking, feasibility without offline pre-training, and algorithmic transparency, we propose a new single object tracking method, called the green object tracker (GOT), in this work. GOT conducts an ensemble of three prediction branches for robust box tracking: 1) a global object-based correlator to predict the object location roughly, 2) a local patch-based correlator to build temporal correlations of small spatial units, and 3) a superpixel-based segmentator to exploit the spatial information of the target frame. GOT offers competitive tracking accuracy with state-of-the-art unsupervised trackers, which demand heavy offline pre-training, at a lower computation cost. GOT has a tiny model size (<3k parameters) and low inference complexity (around 58M FLOPs per frame). Since its inference complexity is between 0.1%-10% of DL trackers, it can be easily deployed on mobile and edge devices.

Hong-Shuo Chen, Shuowen Hu, Suya You et al.

On the basis of DefakeHop, an enhanced lightweight Deepfake detector called DefakeHop++ is proposed in this work. The improvements lie in two areas. First, DefakeHop examines three facial regions (i.e., two eyes and mouth) while DefakeHop++ includes eight more landmarks for broader coverage. Second, for discriminant features selection, DefakeHop uses an unsupervised approach while DefakeHop++ adopts a more effective approach with supervision, called the Discriminant Feature Test (DFT). In DefakeHop++, rich spatial and spectral features are first derived from facial regions and landmarks automatically. Then, DFT is used to select a subset of discriminant features for classifier training. As compared with MobileNet v3 (a lightweight CNN model of 1.5M parameters targeting at mobile applications), DefakeHop++ has a model of 238K parameters, which is 16% of MobileNet v3. Furthermore, DefakeHop++ outperforms MobileNet v3 in Deepfake image detection performance in a weakly-supervised setting.

Zhiruo Zhou, Hongyu Fu, Suya You et al.

Supervised and unsupervised deep trackers that rely on deep learning technologies are popular in recent years. Yet, they demand high computational complexity and a high memory cost. A green unsupervised single-object tracker, called GUSOT, that aims at object tracking for long videos under a resource-constrained environment is proposed in this work. Built upon a baseline tracker, UHP-SOT++, which works well for short-term tracking, GUSOT contains two additional new modules: 1) lost object recovery, and 2) color-saliency-based shape proposal. They help resolve the tracking loss problem and offer a more flexible object proposal, respectively. Thus, they enable GUSOT to achieve higher tracking accuracy in the long run. We conduct experiments on the large-scale dataset LaSOT with long video sequences, and show that GUSOT offers a lightweight high-performance tracking solution that finds applications in mobile and edge computing platforms.

Ganning Zhao, Tingwei Shen, Suya You et al.

Ensuring the realism of computer-generated synthetic images is crucial to deep neural network (DNN) training. Due to different semantic distributions between synthetic and real-world captured datasets, there exists semantic mismatch between synthetic and refined images, which in turn results in the semantic distortion. Recently, contrastive learning (CL) has been successfully used to pull correlated patches together and push uncorrelated ones apart. In this work, we exploit semantic and structural consistency between synthetic and refined images and adopt CL to reduce the semantic distortion. Besides, we incorporate hard negative mining to improve the performance furthermore. We compare the performance of our method with several other benchmarking methods using qualitative and quantitative measures and show that our method offers the state-of-the-art performance.

A S M Iftekhar, Raphael Ruschel, Satish Kumar et al.

Scene-graph generation involves creating a structural representation of the relationships between objects in a scene by predicting subject-object-relation triplets from input data. Existing methods show poor performance in detecting triplets outside of a predefined set, primarily due to their reliance on dependent feature learning. To address this issue, we propose DDS -- a decoupled dynamic scene-graph generation network -- that consists of two independent branches that can disentangle extracted features. The key innovation of the current paper is the decoupling of the features representing the relationships from those of the objects, which enables the detection of novel object-relationship combinations. The DDS model is evaluated on three datasets and outperforms previous methods by a significant margin, especially in detecting previously unseen triplets.

Zifan Yu, Suya You, Fengbo Ren

Frequency-domain learning draws attention due to its superior tradeoff between inference accuracy and input data size. Frequency-domain learning in 2D computer vision tasks has shown that 2D convolutional neural networks (CNN) have a stationary spectral bias towards low-frequency channels so that high-frequency channels can be pruned with no or little accuracy degradation. However, frequency-domain learning has not been studied in the context of 3D CNNs with 3D volumetric data. In this paper, we study frequency-domain learning for volumetric-based 3D data perception to reveal the spectral bias and the accuracy-input-data-size tradeoff of 3D CNNs. Our study finds that 3D CNNs are sensitive to a limited number of critical frequency channels, especially low-frequency channels. Experiment results show that frequency-domain learning can significantly reduce the size of volumetric-based 3D inputs (based on spectral bias) while achieving comparable accuracy with conventional spatial-domain learning approaches. Specifically, frequency-domain learning is able to reduce the input data size by 98% in 3D shape classification while limiting the average accuracy drop within 2%, and by 98% in the 3D point cloud semantic segmentation with a 1.48% mean-class accuracy improvement while limiting the mean-class IoU loss within 1.55%. Moreover, by learning from higher-resolution 3D data (i.e., 2x of the original image in the spatial domain), frequency-domain learning improves the mean-class accuracy and mean-class IoU by 3.04% and 0.63%, respectively, while achieving an 87.5% input data size reduction in 3D point cloud semantic segmentation.

Zifan Yu, Meida Chen, Zhikang Zhang et al. · amazon-science

Common image-based LiDAR point cloud semantic segmentation (LiDAR PCSS) approaches have bottlenecks resulting from the boundary-blurring problem of convolution neural networks (CNNs) and quantitation loss of spherical projection. In this work, we propose a transformer-based plug-and-play uncertain point refiner, i.e., TransUPR, to refine selected uncertain points in a learnable manner, which leads to an improved segmentation performance. Uncertain points are sampled from coarse semantic segmentation results of 2D image segmentation where uncertain points are located close to the object boundaries in the 2D range image representation and 3D spherical projection background points. Following that, the geometry and coarse semantic features of uncertain points are aggregated by neighbor points in 3D space without adding expensive computation and memory footprint. Finally, the transformer-based refiner, which contains four stacked self-attention layers, along with an MLP module, is utilized for uncertain point classification on the concatenated features of self-attention layers. As the proposed refiner is independent of 2D CNNs, our TransUPR can be easily integrated into any existing image-based LiDAR PCSS approaches, e.g., CENet. Our TransUPR with the CENet achieves state-of-the-art performance, i.e., 68.2% mean Intersection over Union (mIoU) on the Semantic KITTI benchmark, which provides a performance improvement of 0.6% on the mIoU compared to the original CENet.

Tsung-Shan Yang, Yun-Cheng Wang, Chengwei Wei et al.

Human-Object Interaction (HOI) detection is a fundamental task in image understanding. While deep-learning-based HOI methods provide high performance in terms of mean Average Precision (mAP), they are computationally expensive and opaque in training and inference processes. An Efficient HOI (EHOI) detector is proposed in this work to strike a good balance between detection performance, inference complexity, and mathematical transparency. EHOI is a two-stage method. In the first stage, it leverages a frozen object detector to localize the objects and extract various features as intermediate outputs. In the second stage, the first-stage outputs predict the interaction type using the XGBoost classifier. Our contributions include the application of error correction codes (ECCs) to encode rare interaction cases, which reduces the model size and the complexity of the XGBoost classifier in the second stage. Additionally, we provide a mathematical formulation of the relabeling and decision-making process. Apart from the architecture, we present qualitative results to explain the functionalities of the feedforward modules. Experimental results demonstrate the advantages of ECC-coded interaction labels and the excellent balance of detection performance and complexity of the proposed EHOI method.

Tingwei Shen, Ganning Zhao, Suya You

Synthetic-to-real data translation using generative adversarial learning has achieved significant success in improving synthetic data. Yet, limited studies focus on deep evaluation and comparison of adversarial training on general-purpose synthetic data for machine learning. This work aims to train and evaluate a synthetic-to-real generative model that transforms the synthetic renderings into more realistic styles on general-purpose datasets conditioned with unlabeled real-world data. Extensive performance evaluation and comparison have been conducted through qualitative and quantitative metrics and a defined downstream perception task.

Adam W. Harley, Yang You, Xinglong Sun et al.

We introduce AllTracker: a model that estimates long-range point tracks by way of estimating the flow field between a query frame and every other frame of a video. Unlike existing point tracking methods, our approach delivers high-resolution and dense (all-pixel) correspondence fields, which can be visualized as flow maps. Unlike existing optical flow methods, our approach corresponds one frame to hundreds of subsequent frames, rather than just the next frame. We develop a new architecture for this task, blending techniques from existing work in optical flow and point tracking: the model performs iterative inference on low-resolution grids of correspondence estimates, propagating information spatially via 2D convolution layers, and propagating information temporally via pixel-aligned attention layers. The model is fast and parameter-efficient (16 million parameters), and delivers state-of-the-art point tracking accuracy at high resolution (i.e., tracking 768x1024 pixels, on a 40G GPU). A benefit of our design is that we can train jointly on optical flow datasets and point tracking datasets, and we find that doing so is crucial for top performance. We provide an extensive ablation study on our architecture details and training recipe, making it clear which details matter most. Our code and model weights are available at https://alltracker.github.io

Raphael Ruschel, A. S. M. Iftekhar, B. S. Manjunath et al.

The increasing complexity of modern deep neural network models and the expanding sizes of datasets necessitate the development of optimized and scalable training methods. In this white paper, we addressed the challenge of efficiently training neural network models using sequences of varying sizes. To address this challenge, we propose a novel training scheme that enables efficient distributed data-parallel training on sequences of different sizes with minimal overhead. By using this scheme we were able to reduce the padding amount by more than 100$x$ while not deleting a single frame, resulting in an overall increased performance on both training time and Recall in our experiments.

Zifan Yu, Erfan Bank Tavakoli, Meida Chen et al.

The area of Video Camouflaged Object Detection (VCOD) presents unique challenges in the field of computer vision due to texture similarities between target objects and their surroundings, as well as irregular motion patterns caused by both objects and camera movement. In this paper, we introduce TokenMotion (TMNet), which employs a transformer-based model to enhance VCOD by extracting motion-guided features using a learnable token selection. Evaluated on the challenging MoCA-Mask dataset, TMNet achieves state-of-the-art performance in VCOD. It outperforms the existing state-of-the-art method by a 12.8% improvement in weighted F-measure, an 8.4% enhancement in S-measure, and a 10.7% boost in mean IoU. The results demonstrate the benefits of utilizing motion-guided features via learnable token selection within a transformer-based framework to tackle the intricate task of VCOD.

Ganning Zhao, Wenhui Cui, Suya You et al.

Unsupervised image-to-image (I2I) translation learns cross-domain image mapping that transfers input from the source domain to output in the target domain while preserving its semantics. One challenge is that different semantic statistics in source and target domains result in content discrepancy known as semantic distortion. To address this problem, a novel I2I method that maintains semantic consistency in translation is proposed and named SemST in this work. SemST reduces semantic distortion by employing contrastive learning and aligning the structural and textural properties of input and output by maximizing their mutual information. Furthermore, a multi-scale approach is introduced to enhance translation performance, thereby enabling the applicability of SemST to domain adaptation in high-resolution images. Experiments show that SemST effectively mitigates semantic distortion and achieves state-of-the-art performance. Also, the application of SemST to domain adaptation (DA) is explored. It is demonstrated by preliminary experiments that SemST can be utilized as a beneficial pre-training for the semantic segmentation task.

Shijie Zhou, Haoran Chang, Sicheng Jiang et al.

3D scene representations have gained immense popularity in recent years. Methods that use Neural Radiance fields are versatile for traditional tasks such as novel view synthesis. In recent times, some work has emerged that aims to extend the functionality of NeRF beyond view synthesis, for semantically aware tasks such as editing and segmentation using 3D feature field distillation from 2D foundation models. However, these methods have two major limitations: (a) they are limited by the rendering speed of NeRF pipelines, and (b) implicitly represented feature fields suffer from continuity artifacts reducing feature quality. Recently, 3D Gaussian Splatting has shown state-of-the-art performance on real-time radiance field rendering. In this work, we go one step further: in addition to radiance field rendering, we enable 3D Gaussian splatting on arbitrary-dimension semantic features via 2D foundation model distillation. This translation is not straightforward: naively incorporating feature fields in the 3DGS framework encounters significant challenges, notably the disparities in spatial resolution and channel consistency between RGB images and feature maps. We propose architectural and training changes to efficiently avert this problem. Our proposed method is general, and our experiments showcase novel view semantic segmentation, language-guided editing and segment anything through learning feature fields from state-of-the-art 2D foundation models such as SAM and CLIP-LSeg. Across experiments, our distillation method is able to provide comparable or better results, while being significantly faster to both train and render. Additionally, to the best of our knowledge, we are the first method to enable point and bounding-box prompting for radiance field manipulation, by leveraging the SAM model. Project website at: https://feature-3dgs.github.io/

Masud Ahmed, Zahid Hasan, Syed Arefinul Haque et al.

Traditional transformer-based semantic segmentation relies on quantized embeddings. However, our analysis reveals that autoencoder accuracy on segmentation mask using quantized embeddings (e.g. VQ-VAE) is 8% lower than continuous-valued embeddings (e.g. KL-VAE). Motivated by this, we propose a continuous-valued embedding framework for semantic segmentation. By reformulating semantic mask generation as a continuous image-to-embedding diffusion process, our approach eliminates the need for discrete latent representations while preserving fine-grained spatial and semantic details. Our key contribution includes a diffusion-guided autoregressive transformer that learns a continuous semantic embedding space by modeling long-range dependencies in image features. Our framework contains a unified architecture combining a VAE encoder for continuous feature extraction, a diffusion-guided transformer for conditioned embedding generation, and a VAE decoder for semantic mask reconstruction. Our setting facilitates zero-shot domain adaptation capabilities enabled by the continuity of the embedding space. Experiments across diverse datasets (e.g., Cityscapes and domain-shifted variants) demonstrate state-of-the-art robustness to distribution shifts, including adverse weather (e.g., fog, snow) and viewpoint variations. Our model also exhibits strong noise resilience, achieving robust performance ($\approx$ 95% AP compared to baseline) under gaussian noise, moderate motion blur, and moderate brightness/contrast variations, while experiencing only a moderate impact ($\approx$ 90% AP compared to baseline) from 50% salt and pepper noise, saturation and hue shifts. Code available: https://github.com/mahmed10/CAMSS.git

Yiqi Zhong, Cho-Ying Wu, Suya You et al.

In this paper, we propose our Correlation For Completion Network (CFCNet), an end-to-end deep learning model that uses the correlation between two data sources to perform sparse depth completion. CFCNet learns to capture, to the largest extent, the semantically correlated features between RGB and depth information. Through pairs of image pixels and the visible measurements in a sparse depth map, CFCNet facilitates feature-level mutual transformation of different data sources. Such a transformation enables CFCNet to predict features and reconstruct data of missing depth measurements according to their corresponding, transformed RGB features. We extend canonical correlation analysis to a 2D domain and formulate it as one of our training objectives (i.e. 2d deep canonical correlation, or "2D2CCA loss"). Extensive experiments validate the ability and flexibility of our CFCNet compared to the state-of-the-art methods on both indoor and outdoor scenes with different real-life sparse patterns. Codes are available at: https://github.com/choyingw/CFCNet.

Shijie Zhou, Zhiwen Fan, Dejia Xu et al.

The increasing demand for virtual reality applications has highlighted the significance of crafting immersive 3D assets. We present a text-to-3D 360$^{\circ}$ scene generation pipeline that facilitates the creation of comprehensive 360$^{\circ}$ scenes for in-the-wild environments in a matter of minutes. Our approach utilizes the generative power of a 2D diffusion model and prompt self-refinement to create a high-quality and globally coherent panoramic image. This image acts as a preliminary "flat" (2D) scene representation. Subsequently, it is lifted into 3D Gaussians, employing splatting techniques to enable real-time exploration. To produce consistent 3D geometry, our pipeline constructs a spatially coherent structure by aligning the 2D monocular depth into a globally optimized point cloud. This point cloud serves as the initial state for the centroids of 3D Gaussians. In order to address invisible issues inherent in single-view inputs, we impose semantic and geometric constraints on both synthesized and input camera views as regularizations. These guide the optimization of Gaussians, aiding in the reconstruction of unseen regions. In summary, our method offers a globally consistent 3D scene within a 360$^{\circ}$ perspective, providing an enhanced immersive experience over existing techniques. Project website at: http://dreamscene360.github.io/

Shijie Zhou, Hui Ren, Yijia Weng et al.

Recent advancements in 2D and multimodal models have achieved remarkable success by leveraging large-scale training on extensive datasets. However, extending these achievements to enable free-form interactions and high-level semantic operations with complex 3D/4D scenes remains challenging. This difficulty stems from the limited availability of large-scale, annotated 3D/4D or multi-view datasets, which are crucial for generalizable vision and language tasks such as open-vocabulary and prompt-based segmentation, language-guided editing, and visual question answering (VQA). In this paper, we introduce Feature4X, a universal framework designed to extend any functionality from 2D vision foundation model into the 4D realm, using only monocular video input, which is widely available from user-generated content. The "X" in Feature4X represents its versatility, enabling any task through adaptable, model-conditioned 4D feature field distillation. At the core of our framework is a dynamic optimization strategy that unifies multiple model capabilities into a single representation. Additionally, to the best of our knowledge, Feature4X is the first method to distill and lift the features of video foundation models (e.g., SAM2, InternVideo2) into an explicit 4D feature field using Gaussian Splatting. Our experiments showcase novel view segment anything, geometric and appearance scene editing, and free-form VQA across all time steps, empowered by LLMs in feedback loops. These advancements broaden the scope of agentic AI applications by providing a foundation for scalable, contextually and spatiotemporally aware systems capable of immersive dynamic 4D scene interaction.

Jintang Xue, Ganning Zhao, Jie-En Yao et al.

Understanding 3D scenes goes beyond simply recognizing objects; it requires reasoning about the spatial and semantic relationships between them. Current 3D scene-language models often struggle with this relational understanding, particularly when visual embeddings alone do not adequately convey the roles and interactions of objects. In this paper, we introduce Descrip3D, a novel and powerful framework that explicitly encodes the relationships between objects using natural language. Unlike previous methods that rely only on 2D and 3D embeddings, Descrip3D enhances each object with a textual description that captures both its intrinsic attributes and contextual relationships. These relational cues are incorporated into the model through a dual-level integration: embedding fusion and prompt-level injection. This allows for unified reasoning across various tasks such as grounding, captioning, and question answering, all without the need for task-specific heads or additional supervision. When evaluated on five benchmark datasets, including ScanRefer, Multi3DRefer, ScanQA, SQA3D, and Scan2Cap, Descrip3D consistently outperforms strong baseline models, demonstrating the effectiveness of language-guided relational representation for understanding complex indoor scenes.

Xinyu Wang, Hong-Shuo Chen, Zhiruo Zhou et al.

Camouflaged object detection (COD) aims to distinguish hidden objects embedded in an environment highly similar to the object. Conventional video-based COD (VCOD) methods explicitly extract motion cues or employ complex deep learning networks to handle the temporal information, which is limited by high complexity and unstable performance. In this work, we propose a green VCOD method named GreenVCOD. Built upon a green ICOD method, GreenVCOD uses long- and short-term temporal neighborhoods (TN) to capture joint spatial/temporal context information for decision refinement. Experimental results show that GreenVCOD offers competitive performance compared to state-of-the-art VCOD benchmarks.

Raphael Ruschel, Md Awsafur Rahman, Hardik Prajapati et al.

Understanding video content is pivotal for advancing real-world applications like activity recognition, autonomous systems, and human-computer interaction. While scene graphs are adept at capturing spatial relationships between objects in individual frames, extending these representations to capture dynamic interactions across video sequences remains a significant challenge. To address this, we present TCDSG, Temporally Consistent Dynamic Scene Graphs, an innovative end-to-end framework that detects, tracks, and links subject-object relationships across time, generating action tracklets, temporally consistent sequences of entities and their interactions. Our approach leverages a novel bipartite matching mechanism, enhanced by adaptive decoder queries and feedback loops, ensuring temporal coherence and robust tracking over extended sequences. This method not only establishes a new benchmark by achieving over 60% improvement in temporal recall@k on the Action Genome, OpenPVSG, and MEVA datasets but also pioneers the augmentation of MEVA with persistent object ID annotations for comprehensive tracklet generation. By seamlessly integrating spatial and temporal dynamics, our work sets a new standard in multi-frame video analysis, opening new avenues for high-impact applications in surveillance, autonomous navigation, and beyond.

Zhiruo Zhou, Hongyu Fu, Suya You et al.

An unsupervised, lightweight and high-performance single object tracker, called UHP-SOT, was proposed by Zhou et al. recently. As an extension, we present an enhanced version and name it UHP-SOT++ in this work. Built upon the foundation of the discriminative-correlation-filters-based (DCF-based) tracker, two new ingredients are introduced in UHP-SOT and UHP-SOT++: 1) background motion modeling and 2) object box trajectory modeling. The main difference between UHP-SOT and UHP-SOT++ is the fusion strategy of proposals from three models (i.e., DCF, background motion and object box trajectory models). An improved fusion strategy is adopted by UHP-SOT++ for more robust tracking performance against large-scale tracking datasets. Our second contribution lies in an extensive evaluation of the performance of state-of-the-art supervised and unsupervised methods by testing them on four SOT benchmark datasets - OTB2015, TC128, UAV123 and LaSOT. Experiments show that UHP-SOT++ outperforms all previous unsupervised methods and several deep-learning (DL) methods in tracking accuracy. Since UHP-SOT++ has extremely small model size, high tracking performance, and low computational complexity (operating at a rate of 20 FPS on an i5 CPU even without code optimization), it is an ideal solution in real-time object tracking on resource-limited platforms. Based on the experimental results, we compare pros and cons of supervised and unsupervised trackers and provide a new perspective to understand the performance gap between supervised and unsupervised methods, which is the third contribution of this work.

Hong-Shuo Chen, Kaitai Zhang, Shuowen Hu et al.

A robust fake satellite image detection method, called Geo-DefakeHop, is proposed in this work. Geo-DefakeHop is developed based on the parallel subspace learning (PSL) methodology. PSL maps the input image space into several feature subspaces using multiple filter banks. By exploring response differences of different channels between real and fake images for a filter bank, Geo-DefakeHop learns the most discriminant channels and uses their soft decision scores as features. Then, Geo-DefakeHop selects a few discriminant features from each filter bank and ensemble them to make a final binary decision. Geo-DefakeHop offers a light-weight high-performance solution to fake satellite images detection. Its model size is analyzed, which ranges from 0.8 to 62K parameters. Furthermore, it is shown by experimental results that it achieves an F1-score higher than 95\% under various common image manipulations such as resizing, compression and noise corruption.

Vibashan VS, Domenick Poster, Suya You et al.

Object detectors trained on large-scale RGB datasets are being extensively employed in real-world applications. However, these RGB-trained models suffer a performance drop under adverse illumination and lighting conditions. Infrared (IR) cameras are robust under such conditions and can be helpful in real-world applications. Though thermal cameras are widely used for military applications and increasingly for commercial applications, there is a lack of robust algorithms to robustly exploit the thermal imagery due to the limited availability of labeled thermal data. In this work, we aim to enhance the object detection performance in the thermal domain by leveraging the labeled visible domain data in an Unsupervised Domain Adaptation (UDA) setting. We propose an algorithm agnostic meta-learning framework to improve existing UDA methods instead of proposing a new UDA strategy. We achieve this by meta-learning the initial condition of the detector, which facilitates the adaptation process with fine updates without overfitting or getting stuck at local optima. However, meta-learning the initial condition for the detection scenario is computationally heavy due to long and intractable computation graphs. Therefore, we propose an online meta-learning paradigm which performs online updates resulting in a short and tractable computation graph. To this end, we demonstrate the superiority of our method over many baselines in the UDA setting, producing a state-of-the-art thermal detector for the KAIST and DSIAC datasets.

Zhiruo Zhou, Hongyu Fu, Suya You et al.

An unsupervised online object tracking method that exploits both foreground and background correlations is proposed and named UHP-SOT (Unsupervised High-Performance Single Object Tracker) in this work. UHP-SOT consists of three modules: 1) appearance model update, 2) background motion modeling, and 3) trajectory-based box prediction. A state-of-the-art discriminative correlation filters (DCF) based tracker is adopted by UHP-SOT as the first module. We point out shortcomings of using the first module alone such as failure in recovering from tracking loss and inflexibility in object box adaptation and then propose the second and third modules to overcome them. Both are novel in single object tracking (SOT). We test UHP-SOT on two popular object tracking benchmarks, TB-50 and TB-100, and show that it outperforms all previous unsupervised SOT methods, achieves a performance comparable with the best supervised deep-learning-based SOT methods, and operates at a fast speed (i.e. 22.7-32.0 FPS on a CPU).

A S M Iftekhar, Satish Kumar, R. Austin McEver et al.

The human-object interaction (HOI) detection task refers to localizing humans, localizing objects, and predicting the interactions between each human-object pair. HOI is considered one of the fundamental steps in truly understanding complex visual scenes. For detecting HOI, it is important to utilize relative spatial configurations and object semantics to find salient spatial regions of images that highlight the interactions between human object pairs. This issue is addressed by the novel self-attention based guided transformer network, GTNet. GTNet encodes this spatial contextual information in human and object visual features via self-attention while achieving state of the art results on both the V-COCO and HICO-DET datasets. Code will be made available online.

Jiesi Hu, Ganning Zhao, Suya You et al.

Our goal is to develop stable, accurate, and robust semantic scene understanding methods for wide-area scene perception and understanding, especially in challenging outdoor environments. To achieve this, we are exploring and evaluating a range of related technology and solutions, including AI-driven multimodal scene perception, fusion, processing, and understanding. This work reports our efforts on the evaluation of a state-of-the-art approach for semantic segmentation with multiple RGB and depth sensing data. We employ four large datasets composed of diverse urban and terrain scenes and design various experimental methods and metrics. In addition, we also develop new strategies of multi-datasets learning to improve the detection and recognition of unseen objects. Extensive experiments, implementations, and results are reported in the paper.

Ganning Zhao, Jiesi Hu, Suya You et al.

Current perception systems often carry multimodal imagers and sensors such as 2D cameras and 3D LiDAR sensors. To fuse and utilize the data for downstream perception tasks, robust and accurate calibration of the multimodal sensor data is essential. We propose a novel deep learning-driven technique (CalibDNN) for accurate calibration among multimodal sensor, specifically LiDAR-Camera pairs. The key innovation of the proposed work is that it does not require any specific calibration targets or hardware assistants, and the entire processing is fully automatic with a single model and single iteration. Results comparison among different methods and extensive experiments on different datasets demonstrates the state-of-the-art performance.

Hong-Shuo Chen, Mozhdeh Rouhsedaghat, Hamza Ghani et al.

A light-weight high-performance Deepfake detection method, called DefakeHop, is proposed in this work. State-of-the-art Deepfake detection methods are built upon deep neural networks. DefakeHop extracts features automatically using the successive subspace learning (SSL) principle from various parts of face images. The features are extracted by c/w Saab transform and further processed by our feature distillation module using spatial dimension reduction and soft classification for each channel to get a more concise description of the face. Extensive experiments are conducted to demonstrate the effectiveness of the proposed DefakeHop method. With a small model size of 42,845 parameters, DefakeHop achieves state-of-the-art performance with the area under the ROC curve (AUC) of 100%, 94.95%, and 90.56% on UADFV, Celeb-DF v1 and Celeb-DF v2 datasets, respectively.

Mozhdeh Rouhsedaghat, Yifan Wang, Shuowen Hu et al.

A non-parametric low-resolution face recognition model for resource-constrained environments with limited networking and computing is proposed in this work. Such environments often demand a small model capable of being effectively trained on a small number of labeled data samples, with low training complexity, and low-resolution input images. To address these challenges, we adopt an emerging explainable machine learning methodology called successive subspace learning (SSL).SSL offers an explainable non-parametric model that flexibly trades the model size for verification performance. Its training complexity is significantly lower since its model is trained in a one-pass feedforward manner without backpropagation. Furthermore, active learning can be conveniently incorporated to reduce the labeling cost. The effectiveness of the proposed model is demonstrated by experiments on the LFW and the CMU Multi-PIE datasets.

Ruiyuan Lin, Suya You, Raghuveer Rao et al.

The construction of a multilayer perceptron (MLP) as a piecewise low-order polynomial approximator using a signal processing approach is presented in this work. The constructed MLP contains one input, one intermediate and one output layers. Its construction includes the specification of neuron numbers and all filter weights. Through the construction, a one-to-one correspondence between the approximation of an MLP and that of a piecewise low-order polynomial is established. Comparison between piecewise polynomial and MLP approximations is made. Since the approximation capability of piecewise low-order polynomials is well understood, our findings shed light on the universal approximation capability of an MLP.

Ruiyuan Lin, Zhiruo Zhou, Suya You et al.

A closed-form solution exists in two-class linear discriminant analysis (LDA), which discriminates two Gaussian-distributed classes in a multi-dimensional feature space. In this work, we interpret the multilayer perceptron (MLP) as a generalization of a two-class LDA system so that it can handle an input composed by multiple Gaussian modalities belonging to multiple classes. Besides input layer $l_{in}$ and output layer $l_{out}$, the MLP of interest consists of two intermediate layers, $l_1$ and $l_2$. We propose a feedforward design that has three stages: 1) from $l_{in}$ to $l_1$: half-space partitionings accomplished by multiple parallel LDAs, 2) from $l_1$ to $l_2$: subspace isolation where one Gaussian modality is represented by one neuron, 3) from $l_2$ to $l_{out}$: class-wise subspace mergence, where each Gaussian modality is connected to its target class. Through this process, we present an automatic MLP design that can specify the network architecture (i.e., the layer number and the neuron number at a layer) and all filter weights in a feedforward one-pass fashion. This design can be generalized to an arbitrary distribution by leveraging the Gaussian mixture model (GMM). Experiments are conducted to compare the performance of the traditional backpropagation-based MLP (BP-MLP) and the new feedforward MLP (FF-MLP).

Mozhdeh Rouhsedaghat, Yifan Wang, Xiou Ge et al.

A light-weight low-resolution face gender classification method, called FaceHop, is proposed in this research. We have witnessed rapid progress in face gender classification accuracy due to the adoption of deep learning (DL) technology. Yet, DL-based systems are not suitable for resource-constrained environments with limited networking and computing. FaceHop offers an interpretable non-parametric machine learning solution. It has desired characteristics such as a small model size, a small training data amount, low training complexity, and low-resolution input images. FaceHop is developed with the successive subspace learning (SSL) principle and built upon the foundation of PixelHop++. The effectiveness of the FaceHop method is demonstrated by experiments. For gray-scale face images of resolution $32 \times 32$ in the LFW and the CMU Multi-PIE datasets, FaceHop achieves correct gender classification rates of 94.63% and 95.12% with model sizes of 16.9K and 17.6K parameters, respectively. It outperforms LeNet-5 in classification accuracy while LeNet-5 has a model size of 75.8K parameters.

Zifan Yu, Suya You

Without using extra 3-D data like points cloud or depth images for providing 3-D information, we retrieve the 3-D object information from single monocular images. The high-quality predicted depth images are recovered from single monocular images, and it is fed into the 2-D object proposal network with corresponding monocular images. Most existing deep learning frameworks with two-streams input data always fuse separate data by concatenating or adding, which views every part of a feature map can contribute equally to the whole task. However, when data are noisy, and too much information is redundant, these methods no longer produce predictions or classifications efficiently. In this report, we propose a two-dimensional CCA(canonical correlation analysis) framework to fuse monocular images and corresponding predicted depth images for basic computer vision tasks like image classification and object detection. Firstly, we implemented different structures with one-dimensional CCA and Alexnet to test the performance on the image classification task. And then, we applied one of these structures with 2D-CCA for object detection. During these experiments, we found that our proposed framework behaves better when taking predicted depth images as inputs with the model trained from ground truth depth.

Yueru Chen, Mozhdeh Rouhsedaghat, Suya You et al.

The successive subspace learning (SSL) principle was developed and used to design an interpretable learning model, known as the PixelHop method,for image classification in our prior work. Here, we propose an improved PixelHop method and call it PixelHop++. First, to make the PixelHop model size smaller, we decouple a joint spatial-spectral input tensor to multiple spatial tensors (one for each spectral component) under the spatial-spectral separability assumption and perform the Saab transform in a channel-wise manner, called the channel-wise (c/w) Saab transform.Second, by performing this operation from one hop to another successively, we construct a channel-decomposed feature tree whose leaf nodes contain features of one dimension (1D). Third, these 1D features are ranked according to their cross-entropy values, which allows us to select a subset of discriminant features for image classification. In PixelHop++, one can control the learning model size of fine-granularity,offering a flexible tradeoff between the model size and the classification performance. We demonstrate the flexibility of PixelHop++ on MNIST, Fashion MNIST, and CIFAR-10 three datasets.

Abinaya Manimaran, Thiyagarajan Ramanathan, Suya You et al.

In this work, we study the power of Saak features as an effort towards interpretable deep learning. Being inspired by the operations of convolutional layers of convolutional neural networks, multi-stage Saak transform was proposed. Based on this foundation, we provide an in-depth examination on Saak features, which are coefficients of the Saak transform, by analyzing their properties through visualization and demonstrating their applications in image classification. Being similar to CNN features, Saak features at later stages have larger receptive fields, yet they are obtained in a one-pass feedforward manner without backpropagation. The whole feature extraction process is transparent and is of extremely low complexity. The discriminant power of Saak features is demonstrated, and their classification performance in three well-known datasets (namely, MNIST, CIFAR-10 and STL-10) is shown by experimental results.

Thiyagarajan Ramanathan, Abinaya Manimaran, Suya You et al.

Image classification is vulnerable to adversarial attacks. This work investigates the robustness of Saak transform against adversarial attacks towards high performance image classification. We develop a complete image classification system based on multi-stage Saak transform. In the Saak transform domain, clean and adversarial images demonstrate different distributions at different spectral dimensions. Selection of the spectral dimensions at every stage can be viewed as an automatic denoising process. Motivated by this observation, we carefully design strategies of feature extraction, representation and classification that increase adversarial robustness. The performances with well-known datasets and attacks are demonstrated by extensive experimental evaluations.

Qiangui Huang, Kevin Zhou, Suya You et al.

Many state-of-the-art computer vision algorithms use large scale convolutional neural networks (CNNs) as basic building blocks. These CNNs are known for their huge number of parameters, high redundancy in weights, and tremendous computing resource consumptions. This paper presents a learning algorithm to simplify and speed up these CNNs. Specifically, we introduce a "try-and-learn" algorithm to train pruning agents that remove unnecessary CNN filters in a data-driven way. With the help of a novel reward function, our agents removes a significant number of filters in CNNs while maintaining performance at a desired level. Moreover, this method provides an easy control of the tradeoff between network performance and its scale. Per- formance of our algorithm is validated with comprehensive pruning experiments on several popular CNNs for visual recognition and semantic segmentation tasks.

Weiyue Wang, Qiangui Huang, Suya You et al.

Recent advances in convolutional neural networks have shown promising results in 3D shape completion. But due to GPU memory limitations, these methods can only produce low-resolution outputs. To inpaint 3D models with semantic plausibility and contextual details, we introduce a hybrid framework that combines a 3D Encoder-Decoder Generative Adversarial Network (3D-ED-GAN) and a Long-term Recurrent Convolutional Network (LRCN). The 3D-ED-GAN is a 3D convolutional neural network trained with a generative adversarial paradigm to fill missing 3D data in low-resolution. LRCN adopts a recurrent neural network architecture to minimize GPU memory usage and incorporates an Encoder-Decoder pair into a Long Short-term Memory Network. By handling the 3D model as a sequence of 2D slices, LRCN transforms a coarse 3D shape into a more complete and higher resolution volume. While 3D-ED-GAN captures global contextual structure of the 3D shape, LRCN localizes the fine-grained details. Experimental results on both real-world and synthetic data show reconstructions from corrupted models result in complete and high-resolution 3D objects.

Qiangui Huang, Weiyue Wang, Kevin Zhou et al.

Recurrent neural network (RNN), as a powerful contextual dependency modeling framework, has been widely applied to scene labeling problems. However, this work shows that directly applying traditional RNN architectures, which unfolds a 2D lattice grid into a sequence, is not sufficient to model structure dependencies in images due to the "impact vanishing" problem. First, we give an empirical analysis about the "impact vanishing" problem. Then, a new RNN unit named Recurrent Neural Network with explicit long range conditioning (RNN-ELC) is designed to alleviate this problem. A novel neural network architecture is built for scene labeling tasks where one of the variants of the new RNN unit, Gated Recurrent Unit with Explicit Long-range Conditioning (GRU-ELC), is used to model multi scale contextual dependencies in images. We validate the use of GRU-ELC units with state-of-the-art performance on three standard scene labeling datasets. Comprehensive experiments demonstrate that the new GRU-ELC unit benefits scene labeling problem a lot as it can encode longer contextual dependencies in images more effectively than traditional RNN units.