Interactive Image Retrieval (IIR) aims to retrieve images that are generally similar to the reference image but under the requested text modification. The existing methods usually concatenate or sum the features of image and text simply and roughly, which, however, is difficult to precisely change the local semantics of the image that the text intends to modify. To solve this problem, we propose a Language Guided Local Infiltration (LGLI) system, which fully utilizes the text information and penetrates text features into image features as much as possible. Specifically, we first propose a Language Prompt Visual Localization (LPVL) module to generate a localization mask which explicitly locates the region (semantics) intended to be modified. Then we introduce a Text Infiltration with Local Awareness (TILA) module, which is deployed in the network to precisely modify the reference image and generate image-text infiltrated representation. Extensive experiments on various benchmark databases validate that our method outperforms most state-of-the-art IIR approaches.
This paper is concerned with the matching stability problem across different decoder layers in DEtection TRansformers (DETR). We point out that the unstable matching in DETR is caused by a multi-optimization path problem, which is highlighted by the one-to-one matching design in DETR. To address this problem, we show that the most important design is to use and only use positional metrics (like IOU) to supervise classification scores of positive examples. Under the principle, we propose two simple yet effective modifications by integrating positional metrics to DETR's classification loss and matching cost, named position-supervised loss and position-modulated cost. We verify our methods on several DETR variants. Our methods show consistent improvements over baselines. By integrating our methods with DINO, we achieve 50.4 and 51.5 AP on the COCO detection benchmark using ResNet-50 backbones under 12 epochs and 24 epochs training settings, achieving a new record under the same setting. We achieve 63.8 AP on COCO detection test-dev with a Swin-Large backbone. Our code will be made available at https://github.com/IDEA-Research/Stable-DINO.
Controllable human image generation (HIG) has numerous real-life applications. State-of-the-art solutions, such as ControlNet and T2I-Adapter, introduce an additional learnable branch on top of the frozen pre-trained stable diffusion (SD) model, which can enforce various conditions, including skeleton guidance of HIG. While such a plug-and-play approach is appealing, the inevitable and uncertain conflicts between the original images produced from the frozen SD branch and the given condition incur significant challenges for the learnable branch, which essentially conducts image feature editing for condition enforcement. In this work, we propose a native skeleton-guided diffusion model for controllable HIG called HumanSD. Instead of performing image editing with dual-branch diffusion, we fine-tune the original SD model using a novel heatmap-guided denoising loss. This strategy effectively and efficiently strengthens the given skeleton condition during model training while mitigating the catastrophic forgetting effects. HumanSD is fine-tuned on the assembly of three large-scale human-centric datasets with text-image-pose information, two of which are established in this work. As shown in Figure 1, HumanSD outperforms ControlNet in terms of accurate pose control and image quality, particularly when the given skeleton guidance is sophisticated.
Domain shift degrades the performance of object detection models in practical applications. To alleviate the influence of domain shift, plenty of previous work try to decouple and learn the domain-invariant (common) features from source domains via domain adversarial learning (DAL). However, inspired by causal mechanisms, we find that previous methods ignore the implicit insignificant non-causal factors hidden in the common features. This is mainly due to the single-view nature of DAL. In this work, we present an idea to remove non-causal factors from common features by multi-view adversarial training on source domains, because we observe that such insignificant non-causal factors may still be significant in other latent spaces (views) due to the multi-mode structure of data. To summarize, we propose a Multi-view Adversarial Discriminator (MAD) based domain generalization model, consisting of a Spurious Correlations Generator (SCG) that increases the diversity of source domain by random augmentation and a Multi-View Domain Classifier (MVDC) that maps features to multiple latent spaces, such that the non-causal factors are removed and the domain-invariant features are purified. Extensive experiments on six benchmarks show our MAD obtains state-of-the-art performance.
Real-world images taken by different cameras with different degradation kernels often result in a cross-device domain gap in image super-resolution. A prevalent attempt to this issue is unsupervised domain adaptation (UDA) that needs to access source data. Considering privacy policies or transmission restrictions of data in many practical applications, we propose a SOurce-free image Super-Resolution framework (SOSR) to address this issue, i.e., adapt a model pre-trained on labeled source data to a target domain with only unlabeled target data. SOSR leverages the source model to generate refined pseudo-labels for teacher-student learning. To better utilize the pseudo-labels, this paper proposes a novel wavelet-based augmentation method, named Wavelet Augmentation Transformer (WAT), which can be flexibly incorporated with existing networks, to implicitly produce useful augmented data. WAT learns low-frequency information of varying levels across diverse samples, which is aggregated efficiently via deformable attention. Furthermore, an uncertainty-aware self-training mechanism is proposed to improve the accuracy of pseudo-labels, with inaccurate predictions being rectified by uncertainty estimation. To acquire better SR results and avoid overfitting pseudo-labels, several regularization losses are proposed to constrain the frequency information between target LR and SR images. Experiments show that without accessing source data, SOSR achieves superior results to the state-of-the-art UDA methods.
Whole-body mesh recovery aims to estimate the 3D human body, face, and hands parameters from a single image. It is challenging to perform this task with a single network due to resolution issues, i.e., the face and hands are usually located in extremely small regions. Existing works usually detect hands and faces, enlarge their resolution to feed in a specific network to predict the parameter, and finally fuse the results. While this copy-paste pipeline can capture the fine-grained details of the face and hands, the connections between different parts cannot be easily recovered in late fusion, leading to implausible 3D rotation and unnatural pose. In this work, we propose a one-stage pipeline for expressive whole-body mesh recovery, named OSX, without separate networks for each part. Specifically, we design a Component Aware Transformer (CAT) composed of a global body encoder and a local face/hand decoder. The encoder predicts the body parameters and provides a high-quality feature map for the decoder, which performs a feature-level upsample-crop scheme to extract high-resolution part-specific features and adopt keypoint-guided deformable attention to estimate hand and face precisely. The whole pipeline is simple yet effective without any manual post-processing and naturally avoids implausible prediction. Comprehensive experiments demonstrate the effectiveness of OSX. Lastly, we build a large-scale Upper-Body dataset (UBody) with high-quality 2D and 3D whole-body annotations. It contains persons with partially visible bodies in diverse real-life scenarios to bridge the gap between the basic task and downstream applications.
The success of ChatGPT has recently attracted numerous efforts to replicate it, with instruction-tuning strategies being a key factor in achieving remarkable results. Instruction-tuning not only significantly enhances the model's performance and generalization but also makes the model's generated results more consistent with human speech patterns. However current research rarely studies the impact of different amounts of instruction data on model performance, especially in the real-world use cases. In this paper we explore the performance of large language models based on instruction tuning across different scales of instruction data. An evaluation dataset consisting of 12 major online use cases is constructed in the experiment. With Bloomz-7B1-mt as the base model, the results show that 1) merely increasing the amount of instruction data leads to continuous improvement in tasks such as open-ended generation, 2) in tasks such as math and code, the model performance curve remains quite flat while increasing data size. We further analyze the possible causes of these phenomena and propose potential future research directions such as effectively selecting high-quality training data, scaling base models and training methods specialized for hard tasks. We will release our training and evaluation datasets, as well as model checkpoints.
Controllability, generalizability and efficiency are the major objectives of constructing face avatars represented by neural implicit field. However, existing methods have not managed to accommodate the three requirements simultaneously. They either focus on static portraits, restricting the representation ability to a specific subject, or suffer from substantial computational cost, limiting their flexibility. In this paper, we propose One-shot Talking face Avatar (OTAvatar), which constructs face avatars by a generalized controllable tri-plane rendering solution so that each personalized avatar can be constructed from only one portrait as the reference. Specifically, OTAvatar first inverts a portrait image to a motion-free identity code. Second, the identity code and a motion code are utilized to modulate an efficient CNN to generate a tri-plane formulated volume, which encodes the subject in the desired motion. Finally, volume rendering is employed to generate an image in any view. The core of our solution is a novel decoupling-by-inverting strategy that disentangles identity and motion in the latent code via optimization-based inversion. Benefiting from the efficient tri-plane representation, we achieve controllable rendering of generalized face avatar at $35$ FPS on A100. Experiments show promising performance of cross-identity reenactment on subjects out of the training set and better 3D consistency.
It is expensive and labour-extensive to label the pixel-wise object masks in a video. As a results, the amount of pixel-wise annotations in existing video instance segmentation (VIS) datasets is small, limiting the generalization capability of trained VIS models. An alternative but much cheaper solution is to use bounding boxes to label instances in videos. Inspired by the recent success of box-supervised image instance segmentation, we first adapt the state-of-the-art pixel-supervised VIS models to a box-supervised VIS (BoxVIS) baseline, and observe only slight performance degradation. We consequently propose to improve BoxVIS performance from two aspects. First, we propose a box-center guided spatial-temporal pairwise affinity (STPA) loss to predict instance masks for better spatial and temporal consistency. Second, we collect a larger scale box-annotated VIS dataset (BVISD) by consolidating the videos from current VIS benchmarks and converting images from the COCO dataset to short pseudo video clips. With the proposed BVISD and the STPA loss, our trained BoxVIS model demonstrates promising instance mask prediction performance. Specifically, it achieves 43.2\% and 29.0\% mask AP on the YouTube-VIS 2021 and OVIS valid sets, respectively, exhibiting comparable or even better generalization performance than state-of-the-art pixel-supervised VIS models by using only 16\% annotation time and cost. Codes and data of BoxVIS can be found at \url{https://github.com/MinghanLi/BoxVIS}.
While impressive progress has been achieved, video instance segmentation (VIS) methods with per-clip input often fail on challenging videos with occluded objects and crowded scenes. This is mainly because instance queries in these methods cannot encode well the discriminative embeddings of instances, making the query-based segmenter difficult to distinguish those `hard' instances. To address these issues, we propose to mine discriminative query embeddings (MDQE) to segment occluded instances on challenging videos. First, we initialize the positional embeddings and content features of object queries by considering their spatial contextual information and the inter-frame object motion. Second, we propose an inter-instance mask repulsion loss to distance each instance from its nearby non-target instances. The proposed MDQE is the first VIS method with per-clip input that achieves state-of-the-art results on challenging videos and competitive performance on simple videos. In specific, MDQE with ResNet50 achieves 33.0\% and 44.5\% mask AP on OVIS and YouTube-VIS 2021, respectively. Code of MDQE can be found at \url{https://github.com/MinghanLi/MDQE_CVPR2023}.