ImageNet has been arguably the most popular image classification benchmark, but it is also the one with a significant level of label noise. Recent studies have shown that many samples contain multiple classes, despite being assumed to be a single-label benchmark. They have thus proposed to turn ImageNet evaluation into a multi-label task, with exhaustive multi-label annotations per image. However, they have not fixed the training set, presumably because of a formidable annotation cost. We argue that the mismatch between single-label annotations and effectively multi-label images is equally, if not more, problematic in the training setup, where random crops are applied. With the single-label annotations, a random crop of an image may contain an entirely different object from the ground truth, introducing noisy or even incorrect supervision during training. We thus re-label the ImageNet training set with multi-labels. We address the annotation cost barrier by letting a strong image classifier, trained on an extra source of data, generate the multi-labels. We utilize the pixel-wise multi-label predictions before the final pooling layer, in order to exploit the additional location-specific supervision signals. Training on the re-labeled samples results in improved model performances across the board. ResNet-50 attains the top-1 classification accuracy of 78.9% on ImageNet with our localized multi-labels, which can be further boosted to 80.2% with the CutMix regularization. We show that the models trained with localized multi-labels also outperforms the baselines on transfer learning to object detection and instance segmentation tasks, and various robustness benchmarks. The re-labeled ImageNet training set, pre-trained weights, and the source code are available at {https://github.com/naver-ai/relabel_imagenet}.
We propose an efficient neural network for RAW image denoising. Although neural network-based denoising has been extensively studied for image restoration, little attention has been given to efficient denoising for compute limited and power sensitive devices, such as smartphones and smartwatches. In this paper, we present a novel architecture and a suite of training techniques for high quality denoising in mobile devices. Our work is distinguished by three main contributions. (1) Feature-Align layer that modulates the activations of an encoder-decoder architecture with the input noisy images. The auto modulation layer enforces attention to spatially varying noise that tend to be "washed away" by successive application of convolutions and non-linearity. (2) A novel Feature Matching Loss that allows knowledge distillation from large denoising networks in the form of a perceptual content loss. (3) Empirical analysis of our efficient model trained to specialize on different noise subranges. This opens additional avenue for model size reduction by sacrificing memory for compute. Extensive experimental validation shows that our efficient model produces high quality denoising results that compete with state-of-the-art large networks, while using significantly fewer parameters and MACs. On the Darmstadt Noise Dataset benchmark, we achieve a PSNR of 48.28dB, while using 263 times fewer MACs, and 17.6 times fewer parameters than the state-of-the-art network, which achieves 49.12dB.
One of the fundamental challenges in supervised learning for multimodal image registration is the lack of ground-truth for voxel-level spatial correspondence. This work describes a method to infer voxel-level transformation from higher-level correspondence information contained in anatomical labels. We argue that such labels are more reliable and practical to obtain for reference sets of image pairs than voxel-level correspondence. Typical anatomical labels of interest may include solid organs, vessels, ducts, structure boundaries and other subject-specific ad hoc landmarks. The proposed end-to-end convolutional neural network approach aims to predict displacement fields to align multiple labelled corresponding structures for individual image pairs during the training, while only unlabelled image pairs are used as the network input for inference. We highlight the versatility of the proposed strategy, for training, utilising diverse types of anatomical labels, which need not to be identifiable over all training image pairs. At inference, the resulting 3D deformable image registration algorithm runs in real-time and is fully-automated without requiring any anatomical labels or initialisation. Several network architecture variants are compared for registering T2-weighted magnetic resonance images and 3D transrectal ultrasound images from prostate cancer patients. A median target registration error of 3.6 mm on landmark centroids and a median Dice of 0.87 on prostate glands are achieved from cross-validation experiments, in which 108 pairs of multimodal images from 76 patients were tested with high-quality anatomical labels.
One major impediment in rapidly deploying object detection models for industrial applications is the lack of large annotated datasets. We currently have presented the Sacked Carton Dataset(SCD) that contains carton images from three scenarios such as comprehensive pharmaceutical logistics company(CPLC), e-commerce logistics company(ECLC), fruit market(FM). However, due to domain shift, the model trained with carton datasets from one of the three scenarios in SCD has poor generalization ability when applied to the rest scenarios. To solve this problem, a novel image synthesis method is proposed to replace the foreground texture of the source datasets with the foreground instance texture of the target datasets. This method can greatly augment the target datasets and improve the model's performance. We firstly propose a surfaces segmentation algorithm to identify the different surfaces of the carton instance. Secondly, a contour reconstruction algorithm is proposed to solve the problem of occlusion, truncation, and incomplete contour of carton instances. Finally, the Gaussian fusion algorithm is used to fuse the background from the source datasets with the foreground from the target datasets. The novel image synthesis method can largely boost AP by at least $4.3\%\sim6.5\%$ on RetinaNet and $3.4\%\sim6.8\%$ on Faster R-CNN for the target domain. And on the source domain, the performance AP can be improved by $1.7\%\sim2\%$ on RetinaNet and $0.9\%\sim1.5\%$ on Faster R-CNN. Code is available at https://github.com/hustgetlijun/RCAN.
This paper proposes a novel multi-exposure image fusion method based on exposure compensation. Multi-exposure image fusion is a method to produce images without color saturation regions, by using photos with different exposures. However, in conventional works, it is unclear how to determine appropriate exposure values, and moreover, it is difficult to set appropriate exposure values at the time of photographing due to time constraints. In the proposed method, the luminance of the input multi-exposure images is adjusted on the basis of the relationship between exposure values and pixel values, where the relationship is obtained by assuming that a digital camera has a linear response function. The use of a local contrast enhancement method is also considered to improve input multi-exposure images. The compensated images are finally combined by one of existing multi-exposure image fusion methods. In some experiments, the effectiveness of the proposed method are evaluated in terms of the tone mapped image quality index, statistical naturalness, and discrete entropy, by comparing the proposed one with conventional ones.
As in other areas of medical image analysis, e.g. semantic segmentation, deep learning is currently driving the development of new approaches for image registration. Multi-scale encoder-decoder network architectures achieve state-of-the-art accuracy on tasks such as intra-patient alignment of abdominal CT or brain MRI registration, especially when additional supervision, such as anatomical labels, is available. The success of these methods relies to a large extent on the outstanding ability of deep CNNs to extract descriptive visual features from the input images. In contrast to conventional methods, the explicit inclusion of geometric information plays only a minor role, if at all. In this work we take a look at an exactly opposite approach by investigating a deep learning framework for registration based solely on geometric features and optimisation. We combine graph convolutions with loopy belief message passing to enable highly accurate 3D point cloud registration. Our experimental validation is conducted on complex key-point graphs of inner lung structures, strongly outperforming dense encoder-decoder networks and other point set registration methods. Our code is publicly available at https://github.com/multimodallearning/deep-geo-reg.
Increasing numbers of patients with disabilities or elderly people with mobility issues often suffer from a pressure ulcer. The affected areas need regular checks, but they have a difficulty in accessing a hospital. Some remote diagnosis systems are being used for them, but there are limitations in checking a patient's status regularly. In this paper, we present a remote medical assistant that can help pressure ulcer management with image processing techniques. The proposed system includes a mobile application with a deep learning model for wound segmentation and analysis. As there are not enough data to train the deep learning model, we make use of a pretrained model from a relevant domain and data augmentation that is appropriate for this task. First of all, an image preprocessing method using bilinear interpolation is used to resize images and normalize the images. Second, for data augmentation, we use rotation, reflection, and a watershed algorithm. Third, we use a pretrained deep learning model generated from skin wound images similar to pressure ulcer images. Finally, we added an attention module that can provide hints on the pressure ulcer image features. The resulting model provides an accuracy of 99.0%, an intersection over union (IoU) of 99.99%, and a dice similarity coefficient (DSC) of 93.4% for pressure ulcer segmentation, which is better than existing results.
This manuscript presents a framework towards automated 3D digital documentation and progress reporting of mechanical pipes in building construction projects, using smartphones. New methods were proposed to determine the video frame rate required to achieve a desired image overlap; define metric scale for 3D reconstruction; extract pipes from point clouds; and classify pipes according to their planned bill of quantity radii. The effectiveness of the proposed methods in both laboratory (six pipes) and construction site (58 pipes) conditions was evaluated. It was observed that the proposed metric scale definition achieved sub-millimeter pipe radius estimation accuracy. Both laboratory and field experiments revealed that increasing the image overlap improved the pipe classification quality, radius, and length. Overall, using the proposed methods, it was possible to achieve pipe classification F-measure, radius estimation accuracy, and length estimation percent error of 96.4%, 5.4mm, and 5.0%, respectively, on construction sites using at least 95% image overlap.
Uncertainty quantification (UQ) plays a pivotal role in reduction of uncertainties during both optimization and decision making processes. It can be applied to solve a variety of real-world applications in science and engineering. Bayesian approximation and ensemble learning techniques are two most widely-used UQ methods in the literature. In this regard, researchers have proposed different UQ methods and examined their performance in a variety of applications such as computer vision (e.g., self-driving cars and object detection), image processing (e.g., image restoration), medical image analysis (e.g., medical image classification and segmentation), natural language processing (e.g., text classification, social media texts and recidivism risk-scoring), bioinformatics, etc. This study reviews recent advances in UQ methods used in deep learning. Moreover, we also investigate the application of these methods in reinforcement learning (RL). Then, we outline a few important applications of UQ methods. Finally, we briefly highlight the fundamental research challenges faced by UQ methods and discuss the future research directions in this field.
There is a growing interest in the community in making an embodied AI agent perform a complicated task while interacting with an environment following natural language directives. Recent studies have tackled the problem using ALFRED, a well-designed dataset for the task, but achieved only very low accuracy. This paper proposes a new method, which outperforms the previous methods by a large margin. It is based on a combination of several new ideas. One is a two-stage interpretation of the provided instructions. The method first selects and interprets an instruction without using visual information, yielding a tentative action sequence prediction. It then integrates the prediction with the visual information etc., yielding the final prediction of an action and an object. As the object's class to interact is identified in the first stage, it can accurately select the correct object from the input image. Moreover, our method considers multiple egocentric views of the environment and extracts essential information by applying hierarchical attention conditioned on the current instruction. This contributes to the accurate prediction of actions for navigation. A preliminary version of the method won the ALFRED Challenge 2020. The current version achieves the unseen environment's success rate of 4.45% with a single view, which is further improved to 8.37% with multiple views.