WACV 2024 Papers

Image Style Transfer (IST) is an interdisciplinary topic of computer vision and art that continuously attracts researchers' interests. Different from traditional Image-guided Image Style Transfer (IIST) methods that require a style reference image as input to define the desired style, recent works start to tackle the problem in a text-guided manner, i.e., Text-guided Image Style Transfer (TIST). Compared to IIST, such approaches provide more flexibility with text-specified styles, which are useful in scenarios where the style is hard to define with reference images. Unfortunately, many TIST approaches produce undesirable artifacts in the transferred images. To address this issue, we present a novel method to achieve much improved style transfer based on text guidance. Meanwhile, to offer more flexibility than IIST and TIST, our method allows style inputs from multiple sources and modalities, enabling MultiModality-guided Image Style Transfer (MMIST). Specifically, we realize MMIST with a novel cross-modal GAN inversion method, which generates style representations consistent with specified styles. Such style representations facilitate style transfer and in principle generalize any IIST methods to MMIST. Large-scale experiments and user studies demonstrate that our method achieves state-of-the-art performance on TIST task. Furthermore, comprehensive qualitative results confirm the effectiveness of our method on MMIST task and cross-modal style interpolation.

Stress estimation is key to the early detection and mitigation of health problems, enhancing driving safety through driver stress monitoring, and improving human-robot interaction efficiency by adapting to user's stress levels. In this paper, we present a novel method for video-based remote stress estimation and categorization, which involves two separate experiments: one for stress task classification and another for multilevel stress classification. The method combines two deep learning approaches, the Compact Convolutional Transformer (CCT) and Long Short-Term Memory (LSTM), to form a CCT-LSTM pipeline. For each modality (facial expression and rPPG), a CCT model is used to extract features, followed by an LSTM block for temporal pattern recognition. In stress task classification, T1, T2, and T3 tasks from the UBFC-Phys dataset are used, utilizing sevenfold cross-validation. The results indicated a mean accuracy of 83.2% and an F1 score of 83.4%. For multilevel stress classification, the control (lower stress) and test (higher stress) groups from the same dataset were used with fivefold cross-validation, achieving a mean accuracy of 80.5% and an F1 score of 80.3%. The results suggest that our proposed model surpasses existing stress estimation methods by effectively using multimodal deep learning and the CCT-LSTM pipeline for precise, non-invasive stress detection and categorization, with applications in health monitoring, safety, and interactive technologies.

Recent studies have focused on utilizing multi-modal data to develop robust models for facial Action Unit (AU) detection. However, the heterogeneity of multi-modal data poses challenges in learning effective representations. One such challenge is extracting relevant features from multiple modalities using a single feature extractor. Moreover, previous studies have not fully explored the potential of multi-modal fusion strategies. In contrast to the extensive work on late fusion, there are limited investigations on early fusion for channel information exploration. This paper presents a novel multi-modal reconstruction network, named Multimodal Channel-Mixing (MCM), as a pre-trained model to learn robust representation for facilitating multi-modal fusion. The approach follows an early fusion setup, integrating a Channel-Mixing module, where two out of five channels are randomly dropped. The dropped channels then are reconstructed from the remaining channels using masked autoencoder. This module not only reduces channel redundancy, but also facilitates multi-modal learning and reconstruction capabilities, resulting in robust feature learning. The encoder is fine-tuned on a downstream task of automatic facial action unit detection. Pretraining experiments were conducted on BP4D+, followed by fine-tuning on BP4D and DISFA to assess the effectiveness and robustness of the proposed framework. The results demonstrate that our method meets and surpasses the performance of state-of-the-art baseline method.

Multi-view classification problems are common in medical image analysis, forensics, and other domains where problem queries involve multi-image input. Existing multi-view classification methods are often tailored to a specific task. In this paper, we repurpose off-the-shelf Hybrid CNN-Transformer networks for multi-view classification with either structured or unstructured views. Our approach incorporates a novel fusion scheme, mutual distillation, and introduces minimal additional parameters. We demonstrate the effectiveness and generalization capability of our approach, MV-HFMD, on multiple multi-view classification tasks and show that it outperforms other multi-view approaches, even task-specific methods. Code is available at https://github.com/vidarlab/multi-view-hybrid.

3D object reconstruction is important for semantic scene understanding. It is challenging to reconstruct detailed 3D shapes from monocular images directly due to a lack of depth information, occlusion and noise. Most current methods generate deterministic object models without any awareness of the uncertainty of the reconstruction. We tackle this problem by leveraging a neural object representation which learns an object shape distribution from large dataset of 3d object models and maps it into a latent space. We propose a method to model uncertainty as part of the representation and define an uncertainty-aware encoder which generates latent codes with uncertainty directly from individual input images. Further, we propose a method to propagate the uncertainty in the latent code to SDF values and generate a 3d object mesh with local uncertainty for each mesh component. Finally, we propose an incremental fusion method under a Bayesian framework to fuse the latent codes from multi-view observations. We evaluate the system in both synthetic and real datasets to demonstrate the effectiveness of uncertainty-based fusion to improve 3D object reconstruction accuracy.

Adapting visual object detectors to operational target domains is a challenging task, commonly achieved using unsupervised domain adaptation (UDA) methods. Recent studies have shown that when the labeled dataset comes from multiple source domains, treating them as separate domains and performing a multi-source domain adaptation (MSDA) improves the accuracy and robustness over blending these source domains and performing a UDA. For adaptation, existing MSDA methods learn domain-invariant and domain-specific parameters (for each source domain). However, unlike single-source UDA methods, learning domain-specific parameters makes them grow significantly in proportion to the number of source domains. This paper proposes a novel MSDA method called Prototype-based Mean Teacher (PMT), which uses class prototypes instead of domain-specific subnets to encode domain-specific information. These prototypes are learned using a contrastive loss, aligning the same categories across domains and separating different categories far apart. Given the use of prototypes, the number of parameters required for our PMT method does not increase significantly with the number of source domains, thus reducing memory issues and possible overfitting. Empirical studies indicate that PMT outperforms state-of-the-art MSDA methods on several challenging object detection datasets. Our code is available at https://github.com/imatif17/Prototype-Mean-Teacher

Gaze following estimates gaze targets of in-scene person by understanding human behavior and scene information. Existing methods usually analyze scene images for gaze following. However, compared with visual images, audio also provides crucial cues for determining human behavior.This suggests that we can further improve gaze following considering audio cues. In this paper, we explore gaze following tasks in conversational scenarios. We propose a novel multi-modal gaze following framework based on our observation "audiences tend to focus on the speaker". We first leverage the correlation between audio and lips, and classify speakers and listeners in a scene. We then use the identity information to enhance scene images and propose a gaze candidate estimation network. The network estimates gaze candidates from enhanced scene images and we use MLP to match subjects with candidates as classification tasks. Existing gaze following datasets focus on visual images while ignore audios.To evaluate our method, we collect a conversational dataset, VideoGazeSpeech (VGS), which is the first gaze following dataset including images and audio. Our method significantly outperforms existing methods in VGS datasets. The visualization result also prove the advantage of audio cues in gaze following tasks. Our work will inspire more researches in multi-modal gaze following estimation.

Face relighting is the challenging task of estimating the illumination cast on portrait images by a light source varying in both position and intensity. As shadows are an important aspect of relighting, many prior works focus on estimating accurate shadows using either a shadow mask or face geometry. While these work well, the rendered images do not look aesthetic/photo-realistic. We propose a novel method that combines the features from attention maps at higher resolutions with the lighting information to estimate aesthetic relit images with accurate shadows. We created a new relighting dataset using a synthetic One-Light-At-a-Time (OLAT) lighting rig in Blender software that captures most of the variations encountered in face relighting. Through extensive experimental validation, we show that the performance of our model is better than the current state-of-art face relighting models despite training on a significantly smaller dataset of only synthetic images. We also demonstrate unsupervised domain adaptation from synthetic to real images. We show that our model is able to adapt very well to significantly different out-of-training light source positions.

Semantic segmentation from aerial views is a crucial task for autonomous drones, as they rely on precise and accurate segmentation to navigate safely and efficiently. However, aerial images present unique challenges such as diverse viewpoints, extreme scale variations, and high scene complexity. In this paper, we propose an end-to-end multi-class semantic segmentation diffusion model that addresses these challenges. We introduce recursive denoising to allow information to propagate through the denoising process, as well as a hierarchical multi-scale approach that complements the diffusion process. Our method achieves promising results on the UAVid dataset and state-of-the-art performance on the Vaihingen Building segmentation benchmark. Being the first iteration of this method, it shows great promise for future improvements. Our code and models are available at: https://github.com/benediktkol/recursive-noise-diffusion

Metaverse technologies demand accurate, real-time, and immersive modeling on consumer-grade hardware for both non-human perception (e.g., drone/robot/autonomous car navigation) and immersive technologies like AR/VR, requiring both structural accuracy and photorealism. However, there exists a knowledge gap in how to apply geometric reconstruction and photorealism modeling (novel view synthesis) in a unified framework. To address this gap and promote the development of robust and immersive modeling and rendering with consumer-grade devices, first, we propose a real-world Multi-Sensor Hybrid Room Dataset (MuSHRoom). Our dataset presents exciting challenges and requires state-of-the-art methods to be cost-effective, robust to noisy data and devices, and can jointly learn 3D reconstruction and novel view synthesis, instead of treating them as separate tasks, making them ideal for real-world applications. Second, we benchmark several famous pipelines on our dataset for joint 3D mesh reconstruction and novel view synthesis. Finally, in order to further improve the overall performance, we propose a new method that achieves a good trade-off between the two tasks. Our dataset and benchmark show great potential in promoting the improvements for fusing 3D reconstruction and high-quality rendering in a robust and computationally efficient end-to-end fashion. The dataset and code is available at the project webpate: https://xuqianren.github. io/publications/MuSHRoom/.

Video-based movie genre classification has garnered considerable attention due to its various applications in recommendation systems. Prior work has typically addressed this task by adapting models from traditional video classification tasks, such as action recognition or event detection. However, these models often neglect language elements (e.g., narrations or conversations) present in videos, which can implicitly convey high-level semantics of movie genres, like storylines or background context. Additionally, existing approaches are primarily designed to encode the entire content of the input video, leading to inefficiencies in predicting movie genres. Movie genre prediction may require only a few shots to accurately determine the genres, rendering a comprehensive understanding of the entire video unnecessary. To address these challenges, we propose a Movie genre Classification method based on Language augmentatIon and shot samPling (Movie-CLIP). Movie-CLIP mainly consists of two parts: a language augmentation module to recognize language elements from the input audio, and a shot sampling module to select representative shots from the entire video. We evaluate our method on MovieNet and Condensed Movies datasets, achieving approximate 6-9% improvement in mean Average Precision (mAP) over the baselines. We also generalize Movie-CLIP to the scene boundary detection task, achieving 1.1% improvement in Average Precision (AP) over the state-of-the-art. We release our implementation at github.com/Zhongping-Zhang/Movie-CLIP.

There are numerous applications for human motion synthesis, including animation, gaming, robotics, or sports science. In recent years, human motion generation from natural language has emerged as a promising alternative to costly and labor-intensive data collection methods relying on motion capture or wearable sensors (e.g., suits). Despite this, generating human motion from textual descriptions remains a challenging and intricate task, primarily due to the scarcity of large-scale supervised datasets capable of capturing the full diversity of human activity. This study proposes a new approach, called MotionGPT, to address the limitations of previous text-based human motion generation methods by utilizing the extensive semantic information available in large language models (LLMs). We first pretrain a doubly text-conditional motion diffusion model on both coarse ("high-level") and detailed ("low-level") ground truth text data. Then during inference, we improve motion diversity and alignment with the training set, by zero-shot prompting GPT-3 for additional "low-level" details. Our method achieves new state-of-the-art quantitative results in terms of Frechet Inception Distance (FID) and motion diversity metrics, and improves all considered metrics. Furthermore, it has strong qualitative performance, producing natural results.

Recent transformer-based approaches have demonstrated excellent performance in 3D human pose estimation. However, they have a holistic view and by encoding global relationships between all the joints, they do not capture the local dependencies precisely. In this paper, we present a novel Attention-GCNFormer (AGFormer) block that divides the number of channels by using two parallel transformer and GCNFormer streams. Our proposed GCNFormer module exploits the local relationship between adjacent joints, outputting a new representation that is complementary to the transformer output. By fusing these two representation in an adaptive way, AGFormer exhibits the ability to better learn the underlying 3D structure. By stacking multiple AGFormer blocks, we propose MotionAGFormer in four different variants, which can be chosen based on the speed-accuracy trade-off. We evaluate our model on two popular benchmark datasets: Human3.6M and MPI-INF-3DHP. MotionAGFormer-B achieves state-of-the-art results, with P1 errors of 38.4 mm and 16.2 mm, respectively. Remarkably, it uses a quarter of the parameters and is three times more computationally efficient than the previous leading model on Human3.6M dataset. Code and models are available at https://github.com/TaatiTeam/MotionAGFormer.

Machine learning models for camera-based physiological measurement can have weak generalization due to a lack of representative training data. Body motion is one of the most significant sources of noise when attempting to recover the subtle cardiac pulse from a video. We explore motion transfer as a form of data augmentation to introduce motion variation while preserving physiological changes of interest. We adapt a neural video synthesis approach to augment videos for the task of remote photoplethysmography (rPPG) and study the effects of motion augmentation with respect to 1) the magnitude and 2) the type of motion. After training on motion-augmented versions of publicly available datasets, the presented inter-dataset results on five benchmark datasets show improvements of up to 79% over existing inter-dataset results using TS-CAN, a neural rPPG estimation method. Additionally, we demonstrate a 47% improvement over existing results on the PURE dataset using various state-of-the-art methods. Our findings illustrate the usefulness of motion transfer as a data augmentation technique for improving the generalization of models for camera-based physiological sensing. We release our code for using motion transfer as a data augmentation technique on three publicly available datasets, UBFC-rPPG, PURE, and SCAMPS, and models pre-trained on motion-augmented data.

We propose a novel semi-supervised active learning framework for monocular 3D object detection with LiDAR guidance (MonoLiG), which leverages all modalities of collected data during model development. We utilize LiDAR to guide the data selection and training of monocular 3D detectors without introducing any overhead in the inference phase. During training, we leverage the LiDAR teacher, monocular student cross-modal framework from semi-supervised learning to distill information from unlabeled data as pseudo-labels. To handle the differences in sensor characteristics, we propose a data noise-based weighting mechanism to reduce the effect of propagating noise from LiDAR modality to monocular. For selecting which samples to label to improve the model performance, we propose a sensor consistency-based selection score that is also coherent with the training objective. Extensive experimental results on KITTI and Waymo datasets verify the effectiveness of our proposed framework. In particular, our selection strategy consistently outperforms state-of-the-art active learning baselines, yielding up to 17% better saving rate in labeling costs. Our training strategy attains the top place in KITTI 3D and bird's-eye-view (BEV) monocular object detection official benchmarks by improving the BEV Average Precision (AP) by 2.02. Code is shared at https://github.com/aralhekimoglu/monolig.

Self-supervised monocular depth estimation methods aim to be used in critical applications such as autonomous vehicles for environment analysis. To circumvent the potential imperfections of these approaches, a quantification of the prediction confidence is crucial to guide decision-making systems that rely on depth estimation. In this paper, we propose MonoProb, a new unsupervised monocular depth estimation method that returns an interpretable uncertainty, which means that the uncertainty reflects the expected error of the network in its depth predictions. We rethink the stereo or the structure-from-motion paradigms used to train unsupervised monocular depth models as a probabilistic problem. Within a single forward pass inference, this model provides a depth prediction and a measure of its confidence, without increasing the inference time. We then improve the performance on depth and uncertainty with a novel self-distillation loss for which a student is supervised by a pseudo ground truth that is a probability distribution on depth output by a teacher. To quantify the performance of our models we design new metrics that, unlike traditional ones, measure the absolute performance of uncertainty predictions. Our experiments highlight enhancements achieved by our method on standard depth and uncertainty metrics as well as on our tailored metrics. https://github.com/CEA-LIST/MonoProb

Zero-shot learning offers an efficient solution for a machine learning model to treat unseen categories, avoiding exhaustive data collection. Zero-shot Sketch-based Image Retrieval (ZS-SBIR) simulates real-world scenarios where it is hard and costly to collect paired sketch-photo samples. We propose a novel framework that indirectly aligns sketches and photos by contrasting them through texts, removing the necessity of access to sketch-photo pairs. With an explicit modality encoding learned from data, our approach disentangles modality-agnostic semantics from modality-specific information, bridging the modality gap and enabling effective cross-modal content retrieval within a joint latent space. From comprehensive experiments, we verify the efficacy of the proposed model on ZS-SBIR, and it can be also applied to generalized and fine-grained settings.

Neural video codecs have recently become competitive with standard codecs such as HEVC in the low-delay setting. However, most neural codecs are large floating-point networks that use pixel-dense warping operations for temporal modeling, making them too computationally expensive for deployment on mobile devices. Recent work has demonstrated that running a neural decoder in real time on mobile is feasible, but shows this only for 720p RGB video. This work presents the first neural video codec that decodes 1080p YUV420 video in real time on a mobile device. Our codec relies on two major contributions. First, we design an efficient codec that uses a block-based motion compensation algorithm available on the warping core of the mobile accelerator, and we show how to quantize this model to integer precision. Second, we implement a fast decoder pipeline that concurrently runs neural network components on the neural signal processor, parallel entropy coding on the mobile GPU, and warping on the warping core. Our codec outperforms the previous on-device codec by a large margin with up to 48 % BD-rate savings, while reducing the MAC count on the receiver side by 10x. We perform a careful ablation to demonstrate the effect of the introduced motion compensation scheme, and ablate the effect of model quantization.

We present a system to create Mobile Realistic Fullbody (MoRF) avatars. MoRF avatars are rendered in real-time on mobile devices, learned from monocular videos, and have high realism. We use SMPL-X as a proxy geometry and render it with DNR (neural texture and image-2-image network). We improve on prior work, by overfitting per-frame warping fields in the neural texture space, allowing to better align the training signal between different frames. We also refine SMPL-X mesh fitting procedure to improve the overall avatar quality. In the comparisons to other monocular video-based avatar systems, MoRF avatars achieve higher image sharpness and temporal consistency. Participants of our user study also preferred avatars generated by MoRF.