Scene Text Recognition is a challenging problem because of irregular styles and various distortions. This paper proposed an end-to-end trainable model consists of a finer rectification module and a bidirectional attentional recognition network(Firbarn). The rectification module adopts finer grid to rectify the distorted input image and the bidirectional decoder contains only one decoding layer instead of two separated one. Firbarn can be trained in a weak supervised way, only requiring the scene text images and the corresponding word labels. With the flexible rectification and the novel bidirectional decoder, the results of extensive evaluation on the standard benchmarks show Firbarn outperforms previous works, especially on irregular datasets.
LIDAR point clouds and RGB-images are both extremely essential for 3D object detection. So many state-of-the-art 3D detection algorithms dedicate in fusing these two types of data effectively. However, their fusion methods based on Birds Eye View (BEV) or voxel format are not accurate. In this paper, we propose a novel fusion approach named Point-based Attentive Cont-conv Fusion(PACF) module, which fuses multi-sensor features directly on 3D points. Except for continuous convolution, we additionally add a Point-Pooling and an Attentive Aggregation to make the fused features more expressive. Moreover, based on the PACF module, we propose a 3D multi-sensor multi-task network called Pointcloud-Image RCNN(PI-RCNN as brief), which handles the image segmentation and 3D object detection tasks. PI-RCNN employs a segmentation sub-network to extract full-resolution semantic feature maps from images and then fuses the multi-sensor features via powerful PACF module. Beneficial from the effectiveness of the PACF module and the expressive semantic features from the segmentation module, PI-RCNN can improve much in 3D object detection. We demonstrate the effectiveness of the PACF module and PI-RCNN on the KITTI 3D Detection benchmark, and our method can achieve state-of-the-art on the metric of 3D AP.
Early and accurate prediction of overall survival (OS) time can help to obtain better treatment planning for brain tumor patients. Although many OS time prediction methods have been developed and obtain promising results, there are still several issues. First, conventional prediction methods rely on radiomic features at the local lesion area of a magnetic resonance (MR) volume, which may not represent the full image or model complex tumor patterns. Second, different types of scanners (i.e., multi-modal data) are sensitive to different brain regions, which makes it challenging to effectively exploit the complementary information across multiple modalities and also preserve the modality-specific properties. Third, existing methods focus on prediction models, ignoring complex data-to-label relationships. To address the above issues, we propose an end-to-end OS time prediction model; namely, Multi-modal Multi-channel Network (M2Net). Specifically, we first project the 3D MR volume onto 2D images in different directions, which reduces computational costs, while preserving important information and enabling pre-trained models to be transferred from other tasks. Then, we use a modality-specific network to extract implicit and high-level features from different MR scans. A multi-modal shared network is built to fuse these features using a bilinear pooling model, exploiting their correlations to provide complementary information. Finally, we integrate the outputs from each modality-specific network and the multi-modal shared network to generate the final prediction result. Experimental results demonstrate the superiority of our M2Net model over other methods.
Self-supervised learning by predicting transformations has demonstrated outstanding performances in both unsupervised and (semi-)supervised tasks. Among the state-of-the-art methods is the AutoEncoding Transformations (AET) by decoding transformations from the learned representations of original and transformed images. Both deterministic and probabilistic AETs rely on the Euclidean distance to measure the deviation of estimated transformations from their groundtruth counterparts. However, this assumption is questionable as a group of transformations often reside on a curved manifold rather staying in a flat Euclidean space. For this reason, we should use the geodesic to characterize how an image transform along the manifold of a transformation group, and adopt its length to measure the deviation between transformations. Particularly, we present to autoencode a Lie group of homography transformations PG(2) to learn image representations. For this, we make an estimate of the intractable Riemannian logarithm by projecting PG(2) to a subgroup of rotation transformations SO(3) that allows the closed-form expression of geodesic distances. Experiments demonstrate the proposed AETv2 model outperforms the previous version as well as the other state-of-the-art self-supervised models in multiple tasks.
We develop an edge-assisted object recognition system with the aim of studying the system-level trade-offs between end-to-end latency and object recognition accuracy. We focus on developing techniques that optimize the transmission delay of the system and demonstrate the effect of image encoding rate and neural network size on these two performance metrics. We explore optimal trade-offs between these metrics by measuring the performance of our real time object recognition application. Our measurements reveal hitherto unknown parameter effects and sharp trade-offs, hence paving the road for optimizing this key service. Finally, we formulate two optimization problems using our measurement-based models and following a Pareto analysis we find that careful tuning of the system operation yields at least 33% better performance for real time conditions, over the standard transmission method.
Deep learning-based recommendation models are used pervasively and broadly, for example, to recommend movies, products, or other information most relevant to users, in order to enhance the user experience. Among various application domains which have received significant industry and academia research attention, such as image classification, object detection, language and speech translation, the performance of deep learning-based recommendation models is less well explored, even though recommendation tasks unarguably represent significant AI inference cycles at large-scale datacenter fleets. To advance the state of understanding and enable machine learning system development and optimization for the commerce domain, we aim to define an industry-relevant recommendation benchmark for the MLPerf Training andInference Suites. The paper synthesizes the desirable modeling strategies for personalized recommendation systems. We lay out desirable characteristics of recommendation model architectures and data sets. We then summarize the discussions and advice from the MLPerf Recommendation Advisory Board.
Cancer diagnosis, prognosis, and therapeutic response predictions are based on morphological information from histology slides and molecular profiles from genomic data. However, most deep learning-based objective outcome prediction and grading paradigms are based on histology or genomics alone and do not make use of the complementary information in an intuitive manner. In this work, we propose Pathomic Fusion, a strategy for end-to-end multimodal fusion of histology image and genomic (mutations, CNV, mRNAseq) features for survival outcome prediction. Our approach models pairwise feature interactions across modalities by taking the Kronecker product of gated feature representations and controls the expressiveness of each representation via a gating-based attention mechanism. The proposed framework is able to model pairwise interactions across features in different modalities and control their relative importance. We validate our approach using glioma datasets from the Cancer Genome Atlas (TCGA), which contains paired whole-slide image, genotype, and transcriptome data with ground truth survival and histologic grade labels. Based on a rigorous 15-fold cross-validation, our results demonstrate that the proposed multimodal fusion paradigm improves prognostic determinations from grading and molecular subtyping as well as unimodal deep networks trained on histology and genomic data alone. The proposed method establishes insight and theory on how to train deep networks on multimodal biomedical data in an intuitive manner, which will be useful for other problems in medicine that seek to combine heterogeneous data streams for understanding diseases and predicting response and resistance to treatment.
Automatic tumor segmentation is a crucial step in medical image analysis for computer-aided diagnosis. Although the existing methods based on convolutional neural networks (CNNs) have achieved the state-of-the-art performance, many challenges still remain in medical tumor segmentation. This is because regular CNNs can only exploit translation invariance, ignoring further inherent symmetries existing in medical images such as rotations and reflections. To mitigate this shortcoming, we propose a novel group equivariant segmentation framework by encoding those inherent symmetries for learning more precise representations. First, kernel-based equivariant operations are devised on every orientation, which can effectively address the gaps of learning symmetries in existing approaches. Then, to keep segmentation networks globally equivariant, we design distinctive group layers with layerwise symmetry constraints. By exploiting further symmetries, novel segmentation CNNs can dramatically reduce the sample complexity and the redundancy of filters (by roughly 2/3) over regular CNNs. More importantly, based on our novel framework, we show that a newly built GER-UNet outperforms its regular CNN-based counterpart and the state-of-the-art segmentation methods on real-world clinical data. Specifically, the group layers of our segmentation framework can be seamlessly integrated into any popular CNN-based segmentation architectures.
The most common way for humans to communicate is by speech. But perhaps a language system cannot know what it is communicating without a connection to the real world by image perception. In fact, humans perceive these multiple sources of information together to build a general concept. However, constructing a machine that can alleviate these modalities together in a supervised learning fashion is difficult, because a parallel dataset is required among speech, image, and text modalities altogether that is often unavailable. A machine speech chain based on sequence-to-sequence deep learning was previously proposed to achieve semi-supervised learning that enabled automatic speech recognition (ASR) and text-to-speech synthesis (TTS) to teach each other when they receive unpaired data. In this research, we take a further step by expanding the speech chain into a multimodal chain and design a closely knit chain architecture that connects ASR, TTS, image captioning (IC), and image retrieval (IR) models into a single framework. ASR, TTS, IC, and IR components can be trained in a semi-supervised fashion by assisting each other given incomplete datasets and leveraging cross-modal data augmentation within the chain.
Many recent news headlines have labeled face recognition technology as biased or racist. We report on a methodical investigation into differences in face recognition accuracy between African-American and Caucasian image cohorts of the MORPH dataset. We find that, for all four matchers considered, the impostor and the genuine distributions are statistically significantly different between cohorts. For a fixed decision threshold, the African-American image cohort has a higher false match rate and a lower false non-match rate. ROC curves compare verification rates at the same false match rate, but the different cohorts achieve the same false match rate at different thresholds. This means that ROC comparisons are not relevant to operational scenarios that use a fixed decision threshold. We show that, for the ResNet matcher, the two cohorts have approximately equal separation of impostor and genuine distributions. Using ICAO compliance as a standard of image quality, we find that the initial image cohorts have unequal rates of good quality images. The ICAO-compliant subsets of the original image cohorts show improved accuracy, with the main effect being to reducing the low-similarity tail of the genuine distributions.