Text-to-image (T2I) generative models have recently emerged as a powerful tool, enabling the creation of photo-realistic images and giving rise to a multitude of applications. However, the effective integration of T2I models into fundamental image classification tasks remains an open question. A prevalent strategy to bolster image classification performance is through augmenting the training set with synthetic images generated by T2I models. In this study, we scrutinize the shortcomings of both current generative and conventional data augmentation techniques. Our analysis reveals that these methods struggle to produce images that are both faithful (in terms of foreground objects) and diverse (in terms of background contexts) for domain-specific concepts. To tackle this challenge, we introduce an innovative inter-class data augmentation method known as Diff-Mix (https://github.com/Zhicaiwww/Diff-Mix), which enriches the dataset by performing image translations between classes. Our empirical results demonstrate that Diff-Mix achieves a better balance between faithfulness and diversity, leading to a marked improvement in performance across diverse image classification scenarios, including few-shot, conventional, and long-tail classifications for domain-specific datasets.
We propose a generative model that, given a coarsely edited image, synthesizes a photorealistic output that follows the prescribed layout. Our method transfers fine details from the original image and preserves the identity of its parts. Yet, it adapts it to the lighting and context defined by the new layout. Our key insight is that videos are a powerful source of supervision for this task: objects and camera motions provide many observations of how the world changes with viewpoint, lighting, and physical interactions. We construct an image dataset in which each sample is a pair of source and target frames extracted from the same video at randomly chosen time intervals. We warp the source frame toward the target using two motion models that mimic the expected test-time user edits. We supervise our model to translate the warped image into the ground truth, starting from a pretrained diffusion model. Our model design explicitly enables fine detail transfer from the source frame to the generated image, while closely following the user-specified layout. We show that by using simple segmentations and coarse 2D manipulations, we can synthesize a photorealistic edit faithful to the user's input while addressing second-order effects like harmonizing the lighting and physical interactions between edited objects.
This paper, for the first time, explores text-to-image diffusion models for Zero-Shot Sketch-based Image Retrieval (ZS-SBIR). We highlight a pivotal discovery: the capacity of text-to-image diffusion models to seamlessly bridge the gap between sketches and photos. This proficiency is underpinned by their robust cross-modal capabilities and shape bias, findings that are substantiated through our pilot studies. In order to harness pre-trained diffusion models effectively, we introduce a straightforward yet powerful strategy focused on two key aspects: selecting optimal feature layers and utilising visual and textual prompts. For the former, we identify which layers are most enriched with information and are best suited for the specific retrieval requirements (category-level or fine-grained). Then we employ visual and textual prompts to guide the model's feature extraction process, enabling it to generate more discriminative and contextually relevant cross-modal representations. Extensive experiments on several benchmark datasets validate significant performance improvements.
Neural radiance fields (NeRFs) show potential for transforming images captured worldwide into immersive 3D visual experiences. However, most of this captured visual data remains siloed in our camera rolls as these images contain personal details. Even if made public, the problem of learning 3D representations of billions of scenes captured daily in a centralized manner is computationally intractable. Our approach, DecentNeRF, is the first attempt at decentralized, crowd-sourced NeRFs that require $\sim 10^4\times$ less server computing for a scene than a centralized approach. Instead of sending the raw data, our approach requires users to send a 3D representation, distributing the high computation cost of training centralized NeRFs between the users. It learns photorealistic scene representations by decomposing users' 3D views into personal and global NeRFs and a novel optimally weighted aggregation of only the latter. We validate the advantage of our approach to learn NeRFs with photorealism and minimal server computation cost on structured synthetic and real-world photo tourism datasets. We further analyze how secure aggregation of global NeRFs in DecentNeRF minimizes the undesired reconstruction of personal content by the server.
Vision-and-Language Navigation (VLN) is a challenging task that requires a robot to navigate in photo-realistic environments with human natural language promptings. Recent studies aim to handle this task by constructing the semantic spatial map representation of the environment, and then leveraging the strong ability of reasoning in large language models for generalizing code for guiding the robot navigation. However, these methods face limitations in instance-level and attribute-level navigation tasks as they cannot distinguish different instances of the same object. To address this challenge, we propose a new method, namely, Instance-aware Visual Language Map (IVLMap), to empower the robot with instance-level and attribute-level semantic mapping, where it is autonomously constructed by fusing the RGBD video data collected from the robot agent with special-designed natural language map indexing in the bird's-in-eye view. Such indexing is instance-level and attribute-level. In particular, when integrated with a large language model, IVLMap demonstrates the capability to i) transform natural language into navigation targets with instance and attribute information, enabling precise localization, and ii) accomplish zero-shot end-to-end navigation tasks based on natural language commands. Extensive navigation experiments are conducted. Simulation results illustrate that our method can achieve an average improvement of 14.4\% in navigation accuracy. Code and demo are released at https://ivlmap.github.io/.
Recent works often assume that Vision-Language Model (VLM) representations are based on visual attributes like shape. However, it is unclear to what extent VLMs prioritize this information to represent concepts. We propose Extract and Explore (EX2), a novel approach to characterize important textual features for VLMs. EX2 uses reinforcement learning to align a large language model with VLM preferences and generates descriptions that incorporate the important features for the VLM. Then, we inspect the descriptions to identify the features that contribute to VLM representations. We find that spurious descriptions have a major role in VLM representations despite providing no helpful information, e.g., Click to enlarge photo of CONCEPT. More importantly, among informative descriptions, VLMs rely significantly on non-visual attributes like habitat to represent visual concepts. Also, our analysis reveals that different VLMs prioritize different attributes in their representations. Overall, we show that VLMs do not simply match images to scene descriptions and that non-visual or even spurious descriptions significantly influence their representations.
High Dynamic Range (HDR) content (i.e., images and videos) has a broad range of applications. However, capturing HDR content from real-world scenes is expensive and time-consuming. Therefore, the challenging task of reconstructing visually accurate HDR images from their Low Dynamic Range (LDR) counterparts is gaining attention in the vision research community. A major challenge in this research problem is the lack of datasets, which capture diverse scene conditions (e.g., lighting, shadows, weather, locations, landscapes, objects, humans, buildings) and various image features (e.g., color, contrast, saturation, hue, luminance, brightness, radiance). To address this gap, in this paper, we introduce GTA-HDR, a large-scale synthetic dataset of photo-realistic HDR images sampled from the GTA-V video game. We perform thorough evaluation of the proposed dataset, which demonstrates significant qualitative and quantitative improvements of the state-of-the-art HDR image reconstruction methods. Furthermore, we demonstrate the effectiveness of the proposed dataset and its impact on additional computer vision tasks including 3D human pose estimation, human body part segmentation, and holistic scene segmentation. The dataset, data collection pipeline, and evaluation code are available at: https://github.com/HrishavBakulBarua/GTA-HDR.
Video-driven neural face reenactment aims to synthesize realistic facial images that successfully preserve the identity and appearance of a source face, while transferring the target head pose and facial expressions. Existing GAN-based methods suffer from either distortions and visual artifacts or poor reconstruction quality, i.e., the background and several important appearance details, such as hair style/color, glasses and accessories, are not faithfully reconstructed. Recent advances in Diffusion Probabilistic Models (DPMs) enable the generation of high-quality realistic images. To this end, in this paper we present DiffusionAct, a novel method that leverages the photo-realistic image generation of diffusion models to perform neural face reenactment. Specifically, we propose to control the semantic space of a Diffusion Autoencoder (DiffAE), in order to edit the facial pose of the input images, defined as the head pose orientation and the facial expressions. Our method allows one-shot, self, and cross-subject reenactment, without requiring subject-specific fine-tuning. We compare against state-of-the-art GAN-, StyleGAN2-, and diffusion-based methods, showing better or on-par reenactment performance.
This work presents FlashFace, a practical tool with which users can easily personalize their own photos on the fly by providing one or a few reference face images and a text prompt. Our approach is distinguishable from existing human photo customization methods by higher-fidelity identity preservation and better instruction following, benefiting from two subtle designs. First, we encode the face identity into a series of feature maps instead of one image token as in prior arts, allowing the model to retain more details of the reference faces (e.g., scars, tattoos, and face shape ). Second, we introduce a disentangled integration strategy to balance the text and image guidance during the text-to-image generation process, alleviating the conflict between the reference faces and the text prompts (e.g., personalizing an adult into a "child" or an "elder"). Extensive experimental results demonstrate the effectiveness of our method on various applications, including human image personalization, face swapping under language prompts, making virtual characters into real people, etc. Project Page: https://jshilong.github.io/flashface-page.
Implicit neural representation methods have shown impressive advancements in learning 3D scenes from unstructured in-the-wild photo collections but are still limited by the large computational cost of volumetric rendering. More recently, 3D Gaussian Splatting emerged as a much faster alternative with superior rendering quality and training efficiency, especially for small-scale and object-centric scenarios. Nevertheless, this technique suffers from poor performance on unstructured in-the-wild data. To tackle this, we extend over 3D Gaussian Splatting to handle unstructured image collections. We achieve this by modeling appearance to seize photometric variations in the rendered images. Additionally, we introduce a new mechanism to train transient Gaussians to handle the presence of scene occluders in an unsupervised manner. Experiments on diverse photo collection scenes and multi-pass acquisition of outdoor landmarks show the effectiveness of our method over prior works achieving state-of-the-art results with improved efficiency.