Search

This work proposes FireRedTTS, a foundation text-to-speech framework, to meet the growing demands for personalized and diverse generative speech applications. The framework comprises three parts: data processing, foundation system, and downstream applications. First, we comprehensively present our data processing pipeline, which transforms massive raw audio into a large-scale high-quality TTS dataset with rich annotations and a wide coverage of content, speaking style, and timbre. Then, we propose a language-model-based foundation TTS system. The speech signal is compressed into discrete semantic tokens via a semantic-aware speech tokenizer, and can be generated by a language model from the prompt text and audio. Then, a two-stage waveform generator is proposed to decode them to the high-fidelity waveform. We present two applications of this system: voice cloning for dubbing and human-like speech generation for chatbots. The experimental results demonstrate the solid in-context learning capability of FireRedTTS, which can stably synthesize high-quality speech consistent with the prompt text and audio. For dubbing, FireRedTTS can clone target voices in a zero-shot way for the UGC scenario and adapt to studio-level expressive voice characters in the PUGC scenario via few-shot fine-tuning with 1-hour recording. Moreover, FireRedTTS achieves controllable human-like speech generation in a casual style with paralinguistic behaviors and emotions via instruction tuning, to better serve spoken chatbots.

Deep learning brought boosts to auto diabetic retinopathy (DR) diagnosis, thus, greatly helping ophthalmologists for early disease detection, which contributes to preventing disease deterioration that may eventually lead to blindness. It has been proved that convolutional neural network (CNN)-aided lesion identifying or segmentation benefits auto DR screening. The key to fine-grained lesion tasks mainly lies in: (1) extracting features being both sensitive to tiny lesions and robust against DR-irrelevant interference, and (2) exploiting and re-using encoded information to restore lesion locations under extremely imbalanced data distribution. To this end, we propose a CNN-based DR diagnosis network with attention mechanism involved, termed lesion-aware network, to better capture lesion information from imbalanced data. Specifically, we design the lesion-aware module (LAM) to capture noise-like lesion areas across deeper layers, and the feature-preserve module (FPM) to assist shallow-to-deep feature fusion. Afterward, the proposed lesion-aware network (LANet) is constructed by embedding the LAM and FPM into the CNN decoders for DR-related information utilization. The proposed LANet is then further extended to a DR screening network by adding a classification layer. Through experiments on three public fundus datasets with pixel-level annotations, our method outperforms the mainstream methods with an area under curve of 0.967 in DR screening, and increases the overall average precision by 7.6%, 2.1%, and 1.2% in lesion segmentation on three datasets. Besides, the ablation study validates the effectiveness of the proposed sub-modules., Comment: This is submitted version wihout improvements by reviewers. The final version is published on International Journal of Imaging Systems and Techonology (https://onlinelibrary.wiley.com/doi/10.1002/ima.22933)

In IT system operations, shell commands are common command line tools used by site reliability engineers (SREs) for daily tasks, such as system configuration, package deployment, and performance optimization. The efficiency in their execution has a crucial business impact since shell commands very often aim to execute critical operations, such as the resolution of system faults. However, many shell commands involve long parameters that make them hard to remember and type. Additionally, the experience and knowledge of SREs using these commands are almost always not preserved. In this work, we propose SHREC, a SRE behaviour knowledge graph model for shell command recommendations. We model the SRE shell behaviour knowledge as a knowledge graph and propose a strategy to directly extract such a knowledge from SRE historical shell operations. The knowledge graph is then used to provide shell command recommendations in real-time to improve the SRE operation efficiency. Our empirical study based on real shell commands executed in our company demonstrates that SHREC can improve the SRE operation efficiency, allowing to share and re-utilize the SRE knowledge., Comment: Accepted at IEEE SANER 2024

Image generation can solve insufficient labeled data issues in defect detection. Most defect generation methods are only trained on a single product without considering the consistencies among multiple products, leading to poor quality and diversity of generated results. To address these issues, we propose DefectDiffu, a novel text-guided diffusion method to model both intra-product background consistency and inter-product defect consistency across multiple products and modulate the consistency perturbation directions to control product type and defect strength, achieving diversified defect image generation. Firstly, we leverage a text encoder to separately provide consistency prompts for background, defect, and fusion parts of the disentangled integrated architecture, thereby disentangling defects and normal backgrounds. Secondly, we propose the double-free strategy to generate defect images through two-stage perturbation of consistency direction, thereby controlling product type and defect strength by adjusting the perturbation scale. Besides, DefectDiffu can generate defect mask annotations utilizing cross-attention maps from the defect part. Finally, to improve the generation quality of small defects and masks, we propose the adaptive attention-enhance loss to increase the attention to defects. Experimental results demonstrate that DefectDiffu surpasses state-of-the-art methods in terms of generation quality and diversity, thus effectively improving downstream defection performance. Moreover, defect perturbation directions can be transferred among various products to achieve zero-shot defect generation, which is highly beneficial for addressing insufficient data issues. The code are available at https://github.com/FFDD-diffusion/DefectDiffu.

Reconstructing textureless areas in MVS poses challenges due to the absence of reliable pixel correspondences within fixed patch. Although certain methods employ patch deformation to expand the receptive field, their patches mistakenly skip depth edges to calculate areas with depth discontinuity, thereby causing ambiguity. Consequently, we introduce Multi-granularity Segmentation Prior Multi-View Stereo (MSP-MVS). Specifically, we first propose multi-granularity segmentation prior by integrating multi-granularity depth edges to restrict patch deformation within homogeneous areas. Moreover, we present anchor equidistribution that bring deformed patches with more uniformly distributed anchors to ensure an adequate coverage of their own homogeneous areas. Furthermore, we introduce iterative local search optimization to represent larger patch with sparse representative candidates, significantly boosting the expressive capacity for each patch. The state-of-the-art results on ETH3D and Tanks & Temples benchmarks demonstrate the effectiveness and robust generalization ability of our proposed method., Comment: After a thorough internal review, we identified a significant error in the experimental design described in the Multi-granularity Segmentation Prior Section of our paper, which impacts the accuracy of the data analysis and conclusions. We are in the process of correcting these errors and will submit an updated version in due course

Latest advances have achieved realistic virtual try-on (VTON) through localized garment inpainting using latent diffusion models, significantly enhancing consumers' online shopping experience. However, existing VTON technologies neglect the need for merchants to showcase garments comprehensively, including flexible control over garments, optional faces, poses, and scenes. To address this issue, we define a virtual dressing (VD) task focused on generating freely editable human images with fixed garments and optional conditions. Meanwhile, we design a comprehensive affinity metric index (CAMI) to evaluate the consistency between generated images and reference garments. Then, we propose IMAGDressing-v1, which incorporates a garment UNet that captures semantic features from CLIP and texture features from VAE. We present a hybrid attention module, including a frozen self-attention and a trainable cross-attention, to integrate garment features from the garment UNet into a frozen denoising UNet, ensuring users can control different scenes through text. IMAGDressing-v1 can be combined with other extension plugins, such as ControlNet and IP-Adapter, to enhance the diversity and controllability of generated images. Furthermore, to address the lack of data, we release the interactive garment pairing (IGPair) dataset, containing over 300,000 pairs of clothing and dressed images, and establish a standard pipeline for data assembly. Extensive experiments demonstrate that our IMAGDressing-v1 achieves state-of-the-art human image synthesis performance under various controlled conditions. The code and model will be available at https://github.com/muzishen/IMAGDressing.

Recently, large-scale vision-language models such as CLIP have demonstrated immense potential in zero-shot anomaly segmentation (ZSAS) task, utilizing a unified model to directly detect anomalies on any unseen product with painstakingly crafted text prompts. However, existing methods often assume that the product category to be inspected is known, thus setting product-specific text prompts, which is difficult to achieve in the data privacy scenarios. Moreover, even the same type of product exhibits significant differences due to specific components and variations in the production process, posing significant challenges to the design of text prompts. In this end, we propose a visual context prompting model (VCP-CLIP) for ZSAS task based on CLIP. The insight behind VCP-CLIP is to employ visual context prompting to activate CLIP's anomalous semantic perception ability. In specific, we first design a Pre-VCP module to embed global visual information into the text prompt, thus eliminating the necessity for product-specific prompts. Then, we propose a novel Post-VCP module, that adjusts the text embeddings utilizing the fine-grained features of the images. In extensive experiments conducted on 10 real-world industrial anomaly segmentation datasets, VCP-CLIP achieved state-of-the-art performance in ZSAS task. The code is available at https://github.com/xiaozhen228/VCP-CLIP.

Recent research showcases the considerable potential of conditional diffusion models for generating consistent stories. However, current methods, which predominantly generate stories in an autoregressive and excessively caption-dependent manner, often underrate the contextual consistency and relevance of frames during sequential generation. To address this, we propose a novel Rich-contextual Conditional Diffusion Models (RCDMs), a two-stage approach designed to enhance story generation's semantic consistency and temporal consistency. Specifically, in the first stage, the frame-prior transformer diffusion model is presented to predict the frame semantic embedding of the unknown clip by aligning the semantic correlations between the captions and frames of the known clip. The second stage establishes a robust model with rich contextual conditions, including reference images of the known clip, the predicted frame semantic embedding of the unknown clip, and text embeddings of all captions. By jointly injecting these rich contextual conditions at the image and feature levels, RCDMs can generate semantic and temporal consistency stories. Moreover, RCDMs can generate consistent stories with a single forward inference compared to autoregressive models. Our qualitative and quantitative results demonstrate that our proposed RCDMs outperform in challenging scenarios. The code and model will be available at https://github.com/muzishen/RCDMs.

In the field of portrait video generation, the use of single images to generate portrait videos has become increasingly prevalent. A common approach involves leveraging generative models to enhance adapters for controlled generation. However, control signals (e.g., text, audio, reference image, pose, depth map, etc.) can vary in strength. Among these, weaker conditions often struggle to be effective due to interference from stronger conditions, posing a challenge in balancing these conditions. In our work on portrait video generation, we identified audio signals as particularly weak, often overshadowed by stronger signals such as facial pose and reference image. However, direct training with weak signals often leads to difficulties in convergence. To address this, we propose V-Express, a simple method that balances different control signals through the progressive training and the conditional dropout operation. Our method gradually enables effective control by weak conditions, thereby achieving generation capabilities that simultaneously take into account the facial pose, reference image, and audio. The experimental results demonstrate that our method can effectively generate portrait videos controlled by audio. Furthermore, a potential solution is provided for the simultaneous and effective use of conditions of varying strengths.

The success of the text-guided diffusion model has inspired the development and release of numerous powerful diffusion models within the open-source community. These models are typically fine-tuned on various expert datasets, showcasing diverse denoising capabilities. Leveraging multiple high-quality models to produce stronger generation ability is valuable, but has not been extensively studied. Existing methods primarily adopt parameter merging strategies to produce a new static model. However, they overlook the fact that the divergent denoising capabilities of the models may dynamically change across different states, such as when experiencing different prompts, initial noises, denoising steps, and spatial locations. In this paper, we propose a novel ensembling method, Adaptive Feature Aggregation (AFA), which dynamically adjusts the contributions of multiple models at the feature level according to various states (i.e., prompts, initial noises, denoising steps, and spatial locations), thereby keeping the advantages of multiple diffusion models, while suppressing their disadvantages. Specifically, we design a lightweight Spatial-Aware Block-Wise (SABW) feature aggregator that adaptive aggregates the block-wise intermediate features from multiple U-Net denoisers into a unified one. The core idea lies in dynamically producing an individual attention map for each model's features by comprehensively considering various states. It is worth noting that only SABW is trainable with about 50 million parameters, while other models are frozen. Both the quantitative and qualitative experiments demonstrate the effectiveness of our proposed Adaptive Feature Aggregation method. The code is available at https://github.com/tenvence/afa/.

Remote sensing target detection aims to identify and locate critical targets within remote sensing images, finding extensive applications in agriculture and urban planning. Feature pyramid networks (FPNs) are commonly used to extract multi-scale features. However, existing FPNs often overlook extracting low-level positional information and fine-grained context interaction. To address this, we propose a novel location refined feature pyramid network (LR-FPN) to enhance the extraction of shallow positional information and facilitate fine-grained context interaction. The LR-FPN consists of two primary modules: the shallow position information extraction module (SPIEM) and the contextual interaction module (CIM). Specifically, SPIEM first maximizes the retention of solid location information of the target by simultaneously extracting positional and saliency information from the low-level feature map. Subsequently, CIM injects this robust location information into different layers of the original FPN through spatial and channel interaction, explicitly enhancing the object area. Moreover, in spatial interaction, we introduce a simple local and non-local interaction strategy to learn and retain the saliency information of the object. Lastly, the LR-FPN can be readily integrated into common object detection frameworks to improve performance significantly. Extensive experiments on two large-scale remote sensing datasets (i.e., DOTAV1.0 and HRSC2016) demonstrate that the proposed LR-FPN is superior to state-of-the-art object detection approaches. Our code and models will be publicly available.

With the leaping advances in autonomous vehicles and transportation infrastructure, dual function radar-communication (DFRC) systems have become attractive due to the size, cost and resource efficiency. A frequency modulated continuous waveform (FMCW)-based radar-communication system (FRaC) utilizing both sparse multiple-input and multiple-output (MIMO) arrays and index modulation (IM) has been proposed to form a DFRC system specifically designed for vehicular applications. In this paper, the three-dimensional (3D) parameter estimation problem in the FRaC is considered. Since the 3D-parameters including range, direction of arrival (DOA) and velocity are coupled in the estimating matrix of the FRaC system, the existing estimation algorithms cannot estimate the 3D-parameters accurately. Hence, a novel decomposed decoupled atomic norm minimization (DANM) method is proposed by splitting the 3D-parameter estimating matrix into multiple 2D matrices with sparsity constraints. Then, the 3D-parameters are estimated and efficiently and separately with the optimized decoupled estimating matrix. Moreover, the Cram\'{e}r-Rao lower bound (CRLB) of the 3D-parameter estimation are derived, and the computational complexity of the proposed algorithm is analyzed. Simulation results show that the proposed decomposed DANM method exploits the advantage of the virtual aperture in the existence of coupling caused by IM and sparse MIMO array and outperforms the co-estimation algorithm with lower computation complexity.

Recent work has showcased the significant potential of diffusion models in pose-guided person image synthesis. However, owing to the inconsistency in pose between the source and target images, synthesizing an image with a distinct pose, relying exclusively on the source image and target pose information, remains a formidable challenge. This paper presents Progressive Conditional Diffusion Models (PCDMs) that incrementally bridge the gap between person images under the target and source poses through three stages. Specifically, in the first stage, we design a simple prior conditional diffusion model that predicts the global features of the target image by mining the global alignment relationship between pose coordinates and image appearance. Then, the second stage establishes a dense correspondence between the source and target images using the global features from the previous stage, and an inpainting conditional diffusion model is proposed to further align and enhance the contextual features, generating a coarse-grained person image. In the third stage, we propose a refining conditional diffusion model to utilize the coarsely generated image from the previous stage as a condition, achieving texture restoration and enhancing fine-detail consistency. The three-stage PCDMs work progressively to generate the final high-quality and high-fidelity synthesized image. Both qualitative and quantitative results demonstrate the consistency and photorealism of our proposed PCDMs under challenging scenarios.The code and model will be available at https://github.com/tencent-ailab/PCDMs., Comment: Accepted to ICLR 2024

In industrial defect segmentation tasks, while pixel accuracy and Intersection over Union (IoU) are commonly employed metrics to assess segmentation performance, the output consistency (also referred to equivalence) of the model is often overlooked. Even a small shift in the input image can yield significant fluctuations in the segmentation results. Existing methodologies primarily focus on data augmentation or anti-aliasing to enhance the network's robustness against translational transformations, but their shift equivalence performs poorly on the test set or is susceptible to nonlinear activation functions. Additionally, the variations in boundaries resulting from the translation of input images are consistently disregarded, thus imposing further limitations on the shift equivalence. In response to this particular challenge, a novel pair of down/upsampling layers called component attention polyphase sampling (CAPS) is proposed as a replacement for the conventional sampling layers in CNNs. To mitigate the effect of image boundary variations on the equivalence, an adaptive windowing module is designed in CAPS to adaptively filter out the border pixels of the image. Furthermore, a component attention module is proposed to fuse all downsampled features to improve the segmentation performance. The experimental results on the micro surface defect (MSD) dataset and four real-world industrial defect datasets demonstrate that the proposed method exhibits higher equivalence and segmentation performance compared to other state-of-the-art methods.Our code will be available at https://github.com/xiaozhen228/CAPS., Comment: submit to IEEE Transactions on Instrumentation & Measurement

One-class classification (OCC) is a longstanding method for anomaly detection. With the powerful representation capability of the pre-trained backbone, OCC methods have witnessed significant performance improvements. Typically, most of these OCC methods employ transfer learning to enhance the discriminative nature of the pre-trained backbone's features, thus achieving remarkable efficacy. While most current approaches emphasize feature transfer strategies, we argue that the optimization objective space within OCC methods could also be an underlying critical factor influencing performance. In this work, we conducted a thorough investigation into the optimization objective of OCC. Through rigorous theoretical analysis and derivation, we unveil a key insights: any space with the suitable norm can serve as an equivalent substitute for the hypersphere center, without relying on the distribution assumption of training samples. Further, we provide guidelines for determining the feasible domain of norms for the OCC optimization objective. This novel insight sparks a simple and data-agnostic deep one-class classification method. Our method is straightforward, with a single 1x1 convolutional layer as a trainable projector and any space with suitable norm as the optimization objective. Extensive experiments validate the reliability and efficacy of our findings and the corresponding methodology, resulting in state-of-the-art performance in both one-class classification and industrial vision anomaly detection and segmentation tasks., Comment: 15 paegs, 10 figures

The Vision Challenge Track 1 for Data-Effificient Defect Detection requires competitors to instance segment 14 industrial inspection datasets in a data-defificient setting. This report introduces the technical details of the team Aoi-overfifitting-Team for this challenge. Our method focuses on the key problem of segmentation quality of defect masks in scenarios with limited training samples. Based on the Hybrid Task Cascade (HTC) instance segmentation algorithm, we connect the transformer backbone (Swin-B) through composite connections inspired by CBNetv2 to enhance the baseline results. Additionally, we propose two model ensemble methods to further enhance the segmentation effect: one incorporates semantic segmentation into instance segmentation, while the other employs multi-instance segmentation fusion algorithms. Finally, using multi-scale training and test-time augmentation (TTA), we achieve an average mAP@0.50:0.95 of more than 48.49% and an average mAR@0.50:0.95 of 66.71% on the test set of the Data Effificient Defect Detection Challenge. The code is available at https://github.com/love6tao/Aoi-overfitting-team

Although existing image anomaly detection methods yield impressive results, they are mostly an offline learning paradigm that requires excessive data pre-collection, limiting their adaptability in industrial scenarios with online streaming data. Online learning-based image anomaly detection methods are more compatible with industrial online streaming data but are rarely noticed. For the first time, this paper presents a fully online learning image anomaly detection method, namely LeMO, learning memory for online image anomaly detection. LeMO leverages learnable memory initialized with orthogonal random noise, eliminating the need for excessive data in memory initialization and circumventing the inefficiencies of offline data collection. Moreover, a contrastive learning-based loss function for anomaly detection is designed to enable online joint optimization of memory and image target-oriented features. The presented method is simple and highly effective. Extensive experiments demonstrate the superior performance of LeMO in the online setting. Additionally, in the offline setting, LeMO is also competitive with the current state-of-the-art methods and achieves excellent performance in few-shot scenarios.

Part feature learning is critical for fine-grained semantic understanding in vehicle re-identification. However, existing approaches directly model part features and global features, which can easily lead to serious gradient vanishing issues due to their unequal feature information and unreliable pseudo-labels for unsupervised vehicle re-identification. To address this problem, in this paper, we propose a simple Triplet Contrastive Representation Learning (TCRL) framework which leverages cluster features to bridge the part features and global features for unsupervised vehicle re-identification. Specifically, TCRL devises three memory banks to store the instance/cluster features and proposes a Proxy Contrastive Loss (PCL) to make contrastive learning between adjacent memory banks, thus presenting the associations between the part and global features as a transition of the part-cluster and cluster-global associations. Since the cluster memory bank copes with all the vehicle features, it can summarize them into a discriminative feature representation. To deeply exploit the instance/cluster information, TCRL proposes two additional loss functions. For the instance-level feature, a Hybrid Contrastive Loss (HCL) re-defines the sample correlations by approaching the positive instance features and pushing the all negative instance features away. For the cluster-level feature, a Weighted Regularization Cluster Contrastive Loss (WRCCL) refines the pseudo labels by penalizing the mislabeled images according to the instance similarity. Extensive experiments show that TCRL outperforms many state-of-the-art unsupervised vehicle re-identification approaches.

Object re-identification method is made up of backbone network, feature aggregation, and loss function. However, most backbone networks lack a special mechanism to handle rich scale variations and mine discriminative feature representations. In this paper, we firstly design a hierarchical similarity graph module (HSGM) to reduce the conflict of backbone and re-identification networks. The designed HSGM builds a rich hierarchical graph to mine the mapping relationships between global-local and local-local. Secondly, we divide the feature map along with the spatial and channel directions in each hierarchical graph. The HSGM applies the spatial features and channel features extracted from different locations as nodes, respectively, and utilizes the similarity scores between nodes to construct spatial and channel similarity graphs. During the learning process of HSGM, we utilize a learnable parameter to re-optimize the importance of each position, as well as evaluate the correlation between different nodes. Thirdly, we develop a novel hierarchical similarity graph network (HSGNet) by embedding the HSGM in the backbone network. Furthermore, HSGM can be easily embedded into backbone networks of any depth to improve object re-identification ability. Finally, extensive experiments on three large-scale object datasets demonstrate that the proposed HSGNet is superior to state-of-the-art object re-identification approaches.

Vision-based pattern identification (such as face, fingerprint, iris etc.) has been successfully applied in human biometrics for a long history. However, dog nose-print authentication is a challenging problem since the lack of a large amount of labeled data. For that, this paper presents our proposed methods for dog nose-print authentication (Re-ID) task in CVPR 2022 pet biometric challenge. First, considering the problem that each class only with few samples in the training set, we propose an automatic offline data augmentation strategy. Then, for the difference in sample styles between the training and test datasets, we employ joint cross-entropy, triplet and pair-wise circle losses function for network optimization. Finally, with multiple models ensembled adopted, our methods achieve 86.67\% AUC on the test set. Codes are available at https://github.com/muzishen/Pet-ReID-IMAG., Comment: 3rd place solution to 2022 Pet Biometric Challenge (CVPRW). The source code and trained models can be obtained at this https://github.com/muzishen/Pet-ReID-IMAG

Transformers are more and more popular in computer vision, which treat an image as a sequence of patches and learn robust global features from the sequence. However, pure transformers are not entirely suitable for vehicle re-identification because vehicle re-identification requires both robust global features and discriminative local features. For that, a graph interactive transformer (GiT) is proposed in this paper. In the macro view, a list of GiT blocks are stacked to build a vehicle re-identification model, in where graphs are to extract discriminative local features within patches and transformers are to extract robust global features among patches. In the micro view, graphs and transformers are in an interactive status, bringing effective cooperation between local and global features. Specifically, one current graph is embedded after the former level's graph and transformer, while the current transform is embedded after the current graph and the former level's transformer. In addition to the interaction between graphs and transforms, the graph is a newly-designed local correction graph, which learns discriminative local features within a patch by exploring nodes' relationships. Extensive experiments on three large-scale vehicle re-identification datasets demonstrate that our GiT method is superior to state-of-the-art vehicle re-identification approaches., Comment: Accepted in IEEE TIP 2023