There is a recent trend in the LiDAR perception field towards unifying multiple tasks in a single strong network with improved performance, as opposed to using separate networks for each task. In this paper, we introduce a new LiDAR multi-task learning paradigm based on the transformer. The proposed LiDARFormer utilizes cross-space global contextual feature information and exploits cross-task synergy to boost the performance of LiDAR perception tasks across multiple large-scale datasets and benchmarks. Our novel transformer-based framework includes a cross-space transformer module that learns attentive features between the 2D dense Bird's Eye View (BEV) and 3D sparse voxel feature maps. Additionally, we propose a transformer decoder for the segmentation task to dynamically adjust the learned features by leveraging the categorical feature representations. Furthermore, we combine the segmentation and detection features in a shared transformer decoder with cross-task attention layers to enhance and integrate the object-level and class-level features. LiDARFormer is evaluated on the large-scale nuScenes and the Waymo Open datasets for both 3D detection and semantic segmentation tasks, and it outperforms all previously published methods on both tasks. Notably, LiDARFormer achieves the state-of-the-art performance of 76.4% L2 mAPH and 74.3% NDS on the challenging Waymo and nuScenes detection benchmarks for a single model LiDAR-only method.
Semantic communication is a new paradigm that exploits deep learning models to enable end-to-end communications processes, and recent studies have shown that it can achieve better noise resiliency compared with traditional communication schemes in a low signal-to-noise (SNR) regime. To achieve multiple access in semantic communication, we propose a deep learning-based multiple access (DeepMA) method by training semantic communication models with the abilities of joint source-channel coding (JSCC) and orthogonal signal modulation. DeepMA is achieved by a DeepMA network (DMANet), which is comprised of several independent encoder-decoder pairs (EDPs), and the DeepMA encoders can encode the input data as mutually orthogonal semantic symbol vectors (SSVs) such that the DeepMA decoders can detect and recover their own target data from a received mixed SSV (MSSV) superposed by multiple SSV components transmitted from different encoders. We describe frameworks of DeepMA in wireless device-to-device (D2D), downlink, and uplink channel multiplexing scenarios, along with the training algorithm. We evaluate the performance of the proposed DeepMA in wireless image transmission tasks and compare its performance with the attention module-based deep JSCC (ADJSCC) method and conventional communication schemes using better portable graphics (BPG) and Low-density parity-check code (LDPC). The results obtained show that the proposed DeepMA can achieve effective, flexible, and privacy-preserving channel multiplexing process, and demonstrate that our proposed DeepMA approach can yield comparable bandwidth efficiency compared with conventional multiple access schemes.
Recent deep-learning-based video compression methods brought coding gains over conventional codecs such as AVC and HEVC. However, learning-based codecs generally require considerable computation time and model complexity. In this paper, we propose a new lightweight hybrid video codec consisting of a conventional video codec(HEVC / VVC), a lossless image codec, and our new restoration network. Precisely, our encoder consists of the conventional video encoder and a lossless image encoder, transmitting a lossy-compressed video bitstream along with a losslessly-compressed reference frame. The decoder is constructed with corresponding video/image decoders and a new restoration network, which enhances the compressed video in two-step processes. In the first step, a network trained with a large video dataset restores the details lost by the conventional encoder. Then, we further boost the video quality with the guidance of a reference image, which is a losslessly compressed video frame. The reference image provides video-specific information, which can be utilized to better restore the details of a compressed video. Experimental results show that the proposed method achieves comparable performance to top-tier methods, even when applied to HEVC. Nevertheless, our method has lower complexity, a faster run time, and can be easily integrated into existing conventional codecs.
Physics-informed neural networks have emerged as an alternative method for solving partial differential equations. However, for complex problems, the training of such networks can still require high-fidelity data which can be expensive to generate. To reduce or even eliminate the dependency on high-fidelity data, we propose a novel multi-fidelity architecture which is based on a feature space shared by the low- and high-fidelity solutions. In the feature space, the projections of the low-fidelity and high-fidelity solutions are adjacent by constraining their relative distance. The feature space is represented with an encoder and its mapping to the original solution space is effected through a decoder. The proposed multi-fidelity approach is validated on forward and inverse problems for steady and unsteady problems described by partial differential equations.
One-to-one (o2o) label assignment plays a key role for transformer based end-to-end detection, and it has been recently introduced in fully convolutional detectors for end-to-end dense detection. However, o2o can degrade the feature learning efficiency due to the limited number of positive samples. Though extra positive samples are introduced to mitigate this issue in recent DETRs, the computation of self- and cross- attentions in the decoder limits its practical application to dense and fully convolutional detectors. In this work, we propose a simple yet effective one-to-few (o2f) label assignment strategy for end-to-end dense detection. Apart from defining one positive and many negative anchors for each object, we define several soft anchors, which serve as positive and negative samples simultaneously. The positive and negative weights of these soft anchors are dynamically adjusted during training so that they can contribute more to ``representation learning'' in the early training stage, and contribute more to ``duplicated prediction removal'' in the later stage. The detector trained in this way can not only learn a strong feature representation but also perform end-to-end dense detection. Experiments on COCO and CrowdHuman datasets demonstrate the effectiveness of the o2f scheme. Code is available at https://github.com/strongwolf/o2f.
Accurate 3D mitochondria instance segmentation in electron microscopy (EM) is a challenging problem and serves as a prerequisite to empirically analyze their distributions and morphology. Most existing approaches employ 3D convolutions to obtain representative features. However, these convolution-based approaches struggle to effectively capture long-range dependencies in the volume mitochondria data, due to their limited local receptive field. To address this, we propose a hybrid encoder-decoder framework based on a split spatio-temporal attention module that efficiently computes spatial and temporal self-attentions in parallel, which are later fused through a deformable convolution. Further, we introduce a semantic foreground-background adversarial loss during training that aids in delineating the region of mitochondria instances from the background clutter. Our extensive experiments on three benchmarks, Lucchi, MitoEM-R and MitoEM-H, reveal the benefits of the proposed contributions achieving state-of-the-art results on all three datasets. Our code and models are available at https://github.com/OmkarThawakar/STT-UNET.
Transformers were initially introduced for natural language processing (NLP) tasks, but fast they were adopted by most deep learning fields, including computer vision. They measure the relationships between pairs of input tokens (words in the case of text strings, parts of images for visual Transformers), termed attention. The cost is exponential with the number of tokens. For image classification, the most common Transformer Architecture uses only the Transformer Encoder in order to transform the various input tokens. However, there are also numerous other applications in which the decoder part of the traditional Transformer Architecture is also used. Here, we first introduce the Attention mechanism (Section 1), and then the Basic Transformer Block including the Vision Transformer (Section 2). Next, we discuss some improvements of visual Transformers to account for small datasets or less computation(Section 3). Finally, we introduce Visual Transformers applied to tasks other than image classification, such as detection, segmentation, generation and training without labels (Section 4) and other domains, such as video or multimodality using text or audio data (Section 5).
Deep learning-based edge detectors heavily rely on pixel-wise labels which are often provided by multiple annotators. Existing methods fuse multiple annotations using a simple voting process, ignoring the inherent ambiguity of edges and labeling bias of annotators. In this paper, we propose a novel uncertainty-aware edge detector (UAED), which employs uncertainty to investigate the subjectivity and ambiguity of diverse annotations. Specifically, we first convert the deterministic label space into a learnable Gaussian distribution, whose variance measures the degree of ambiguity among different annotations. Then we regard the learned variance as the estimated uncertainty of the predicted edge maps, and pixels with higher uncertainty are likely to be hard samples for edge detection. Therefore we design an adaptive weighting loss to emphasize the learning from those pixels with high uncertainty, which helps the network to gradually concentrate on the important pixels. UAED can be combined with various encoder-decoder backbones, and the extensive experiments demonstrate that UAED achieves superior performance consistently across multiple edge detection benchmarks. The source code is available at \url{https://github.com/ZhouCX117/UAED}
Human pose is typically represented by a coordinate vector of body joints or their heatmap embeddings. While easy for data processing, unrealistic pose estimates are admitted due to the lack of dependency modeling between the body joints. In this paper, we present a structured representation, named Pose as Compositional Tokens (PCT), to explore the joint dependency. It represents a pose by M discrete tokens with each characterizing a sub-structure with several interdependent joints. The compositional design enables it to achieve a small reconstruction error at a low cost. Then we cast pose estimation as a classification task. In particular, we learn a classifier to predict the categories of the M tokens from an image. A pre-learned decoder network is used to recover the pose from the tokens without further post-processing. We show that it achieves better or comparable pose estimation results as the existing methods in general scenarios, yet continues to work well when occlusion occurs, which is ubiquitous in practice. The code and models are publicly available at https://github.com/Gengzigang/PCT.
Semantic segmentation is still a challenging task for parsing diverse contexts in different scenes, thus the fixed classifier might not be able to well address varying feature distributions during testing. Different from the mainstream literature where the efficacy of strong backbones and effective decoder heads has been well studied, in this paper, additional contextual hints are instead exploited via learning a context-aware classifier whose content is data-conditioned, decently adapting to different latent distributions. Since only the classifier is dynamically altered, our method is model-agnostic and can be easily applied to generic segmentation models. Notably, with only negligible additional parameters and +2\% inference time, decent performance gain has been achieved on both small and large models with challenging benchmarks, manifesting substantial practical merits brought by our simple yet effective method. The implementation is available at \url{https://github.com/tianzhuotao/CAC}.
Prevalent predictive coding-based video compression methods rely on a heavy encoder to reduce temporal redundancy, which makes it challenging to deploy them on resource-constrained devices. Since the 1970s, distributed source coding theory has indicated that independent encoding and joint decoding with side information (SI) can achieve high-efficient compression of correlated sources. This has inspired a distributed coding architecture aiming at reducing the encoding complexity. However, traditional distributed coding methods suffer from a substantial performance gap to predictive coding ones. Inspired by the great success of learning-based compression, we propose the first end-to-end distributed deep video compression framework to improve the rate-distortion performance. A key ingredient is an effective SI generation module at the decoder, which helps to effectively exploit inter-frame correlations without computation-intensive encoder-side motion estimation and compensation. Experiments show that our method significantly outperforms conventional distributed video coding and H.264. Meanwhile, it enjoys 6-7x encoding speedup against DVC [1] with comparable compression performance. Code is released at https://github.com/Xinjie-Q/Distributed-DVC.
Optimizing accuracy and performance while eliminating hallucinations of open-domain conversational large language models (LLMs) is an open research challenge. A particularly promising direction is to augment and ground LLMs with information from structured sources. This paper introduces Conversational Tables (cTBLS), a three-step architecture to retrieve and generate dialogue responses grounded on retrieved tabular information. cTBLS uses Transformer encoder embeddings for Dense Table Retrieval and obtains up to 125% relative improvement over the retriever in the previous state-of-the-art system on the HyrbiDialogue dataset. cTBLS then uses a shared process between encoder and decoder models to perform a coarse+fine tabular knowledge (e.g., cell) ranking combined with a GPT-3.5 LLM response generator to yield a 2x relative improvement in ROUGE scores. Finally, human evaluators prefer cTBLs +80% of the time (coherency, fluency) and judge informativeness to be 4x better than the previous state-of-the-art.
The goal of downscaling an image is to reduce its resolution to a lower resolution, while maintaining the visual characteristics of the original image. Recent learning-based algorithms have shown great improvement over conventional methods in preserving high-frequency information. However, they continue to suffer from various artifacts due to the lack of true high-low resolution image pairs in their training datasets. In this paper, an unsupervised image downscaler that preserves the high frequency content of the original image based on an autoencoder is presented. Specifically, the image downscaler is obtained by extracting the decoder of the developed autoencoder. Furthermore, we propose a preprocessing step that enables image downscaler to any arbitrary scales. Experimental results on five benchmark datasets reveal the qualitative and quantitative superiority of the proposed method at various scales compared to other methods.
Intent-detection and slot-filling are the two main tasks in natural language understanding. In this study, we propose CTRAN, a novel encoder-decoder CNN-Transformer-based architecture for intent-detection and slot-filling. In the encoder, we use BERT, followed by several convolutional layers, and rearrange the output using window feature sequence. We use stacked Transformer encoders after the window feature sequence. For the intent-detection decoder, we utilize self-attention followed by a linear layer. In the slot-filling decoder, we introduce the aligned Transformer decoder, which utilizes a zero diagonal mask, aligning output tags with input tokens. We apply our network on ATIS and SNIPS, and surpass the current state-of-the-art in slot-filling on both datasets. Furthermore, we incorporate the language model as word embeddings, and show that this strategy yields a better result when compared to the language model as an encoder.
In many binary segmentation tasks, most CNNs-based methods use a U-shape encoder-decoder network as their basic structure. They ignore two key problems when the encoder exchanges information with the decoder: one is the lack of interference control mechanism between them, the other is without considering the disparity of the contributions from different encoder levels. In this work, we propose a simple yet general gated network (GateNet) to tackle them all at once. With the help of multi-level gate units, the valuable context information from the encoder can be selectively transmitted to the decoder. In addition, we design a gated dual branch structure to build the cooperation among the features of different levels and improve the discrimination ability of the network. Furthermore, we introduce a "Fold" operation to improve the atrous convolution and form a novel folded atrous convolution, which can be flexibly embedded in ASPP or DenseASPP to accurately localize foreground objects of various scales. GateNet can be easily generalized to many binary segmentation tasks, including general and specific object segmentation and multi-modal segmentation. Without bells and whistles, our network consistently performs favorably against the state-of-the-art methods under 10 metrics on 33 datasets of 10 binary segmentation tasks.
Autonomous vehicles must be capable of handling the occlusion of the environment to ensure safe and efficient driving. In urban environment, occlusion often arises due to other vehicles obscuring the perception of the ego vehicle. Since the occlusion condition can impact the trajectories of vehicles, the behavior of other vehicles is helpful in making inferences about the occlusion as a remedy for perceptual deficiencies. This paper introduces a novel social occlusion inference approach that learns a mapping from agent trajectories and scene context to an occupancy grid map (OGM) representing the view of ego vehicle. Specially, vectorized features are encoded through the polyline encoder to aggregate features of vectors into features of polylines. A transformer module is then utilized to model the high-order interactions of polylines. Importantly, occlusion queries are proposed to fuse polyline features and generate the OGM without the input of visual modality. To verify the performance of vectorized representation, we design a baseline based on a fully transformer encoder-decoder architecture mapping the OGM with occlusion and historical trajectories information to the ground truth OGM. We evaluate our approach on an unsignalized intersection in the INTERACTION dataset, which outperforms the state-of-the-art results.
Just noticeable difference (JND) refers to the maximum visual change that human eyes cannot perceive, and it has a wide range of applications in multimedia systems. However, most existing JND approaches only focus on a single modality, and rarely consider the complementary effects of multimodal information. In this article, we investigate the JND modeling from an end-to-end homologous multimodal perspective, namely hmJND-Net. Specifically, we explore three important visually sensitive modalities, including saliency, depth, and segmentation. To better utilize homologous multimodal information, we establish an effective fusion method via summation enhancement and subtractive offset, and align homologous multimodal features based on a self-attention driven encoder-decoder paradigm. Extensive experimental results on eight different benchmark datasets validate the superiority of our hmJND-Net over eight representative methods.
Automatic radiology reporting has great clinical potential to relieve radiologists from heavy workloads and improve diagnosis interpretation. Recently, researchers have enhanced data-driven neural networks with medical knowledge graphs to eliminate the severe visual and textual bias in this task. The structures of such graphs are exploited by using the clinical dependencies formed by the disease topic tags via general knowledge and usually do not update during the training process. Consequently, the fixed graphs can not guarantee the most appropriate scope of knowledge and limit the effectiveness. To address the limitation, we propose a knowledge graph with Dynamic structure and nodes to facilitate medical report generation with Contrastive Learning, named DCL. In detail, the fundamental structure of our graph is pre-constructed from general knowledge. Then we explore specific knowledge extracted from the retrieved reports to add additional nodes or redefine their relations in a bottom-up manner. Each image feature is integrated with its very own updated graph before being fed into the decoder module for report generation. Finally, this paper introduces Image-Report Contrastive and Image-Report Matching losses to better represent visual features and textual information. Evaluated on IU-Xray and MIMIC-CXR datasets, our DCL outperforms previous state-of-the-art models on these two benchmarks.
Lung cancer has emerged as a severe disease that threatens human life and health. The precise segmentation of lung regions is a crucial prerequisite for localizing tumors, which can provide accurate information for lung image analysis. In this work, we first propose a lung image segmentation model using the NASNet-Large as an encoder and then followed by a decoder architecture, which is one of the most commonly used architectures in deep learning for image segmentation. The proposed NASNet-Large-decoder architecture can extract high-level information and expand the feature map to recover the segmentation map. To further improve the segmentation results, we propose a post-processing layer to remove the irrelevant portion of the segmentation map. Experimental results show that an accurate segmentation model with 0.92 dice scores outperforms state-of-the-art performance.
Accurate medical image segmentation is critical for early medical diagnosis. Most existing methods are based on U-shape structure and use element-wise addition or concatenation to fuse different level features progressively in decoder. However, both the two operations easily generate plenty of redundant information, which will weaken the complementarity between different level features, resulting in inaccurate localization and blurred edges of lesions. To address this challenge, we propose a general multi-scale in multi-scale subtraction network (M$^{2}$SNet) to finish diverse segmentation from medical image. Specifically, we first design a basic subtraction unit (SU) to produce the difference features between adjacent levels in encoder. Next, we expand the single-scale SU to the intra-layer multi-scale SU, which can provide the decoder with both pixel-level and structure-level difference information. Then, we pyramidally equip the multi-scale SUs at different levels with varying receptive fields, thereby achieving the inter-layer multi-scale feature aggregation and obtaining rich multi-scale difference information. In addition, we build a training-free network ``LossNet'' to comprehensively supervise the task-aware features from bottom layer to top layer, which drives our multi-scale subtraction network to capture the detailed and structural cues simultaneously. Without bells and whistles, our method performs favorably against most state-of-the-art methods under different evaluation metrics on eleven datasets of four different medical image segmentation tasks of diverse image modalities, including color colonoscopy imaging, ultrasound imaging, computed tomography (CT), and optical coherence tomography (OCT). The source code can be available at \url{https://github.com/Xiaoqi-Zhao-DLUT/MSNet}.