Purpose Supervised deep learning in radiology suffers from notorious inherent limitations: 1) It requires large, hand-annotated data sets, 2) It is non-generalizable, and 3) It lacks explainability and intuition. We have recently proposed Reinforcement Learning to address all threes. However, we applied it to images with radiologist eye tracking points, which limits the state-action space. Here we generalize the Deep-Q Learning to a gridworld-based environment, so that only the images and image masks are required. Materials and Methods We trained a Deep Q network on 30 two-dimensional image slices from the BraTS brain tumor database. Each image contained one lesion. We then tested the trained Deep Q network on a separate set of 30 testing set images. For comparison, we also trained and tested a keypoint detection supervised deep learning network for the same set of training / testing images. Results Whereas the supervised approach quickly overfit the training data, and predicably performed poorly on the testing set (11\% accuracy), the Deep-Q learning approach showed progressive improved generalizability to the testing set over training time, reaching 70\% accuracy. Conclusion We have shown a proof-of-principle application of reinforcement learning to radiological images, here using 2D contrast-enhanced MRI brain images with the goal of localizing brain tumors. This represents a generalization of recent work to a gridworld setting, naturally suitable for analyzing medical images.
3D Point clouds are a rich source of information that enjoy growing popularity in the vision community. However, due to the sparsity of their representation, learning models based on large point clouds is still a challenge. In this work, we introduce Graphite, a GRAPH-Induced feaTure Extraction pipeline, a simple yet powerful feature transform and keypoint detector. Graphite enables intensive down-sampling of point clouds with keypoint detection accompanied by a descriptor. We construct a generic graph-based learning scheme to describe point cloud regions and extract salient points. To this end, we take advantage of 6D pose information and metric learning to learn robust descriptions and keypoints across different scans. We Reformulate the 3D keypoint pipeline with graph neural networks which allow efficient processing of the point set while boosting its descriptive power which ultimately results in more accurate 3D registrations. We demonstrate our lightweight descriptor on common 3D descriptor matching and point cloud registration benchmarks and achieve comparable results with the state of the art. Describing 100 patches of a point cloud and detecting their keypoints takes only ~0.018 seconds with our proposed network.
Keypoint detection is an essential building block for many robotic applications like motion capture and pose estimation. Historically, keypoints are detected using uniquely engineered markers such as checkerboards or fiducials. More recently, deep learning methods have been explored as they have the ability to detect user-defined keypoints in a marker-less manner. However, different manually selected keypoints can have uneven performance when it comes to detection and localization. An example of this can be found on symmetric robotic tools where DNN detectors cannot solve the correspondence problem correctly. In this work, we propose a new and autonomous way to define the keypoint locations that overcomes these challenges. The approach involves finding the optimal set of keypoints on robotic manipulators for robust visual detection and localization. Using a robotic simulator as a medium, our algorithm utilizes synthetic data for DNN training, and the proposed algorithm is used to optimize the selection of keypoints through an iterative approach. The results show that when using the optimized keypoints, the detection performance of the DNNs improved significantly. We further use the optimized keypoints for real robotic applications by using domain randomization to bridge the reality gap between the simulator and the physical world. The physical world experiments show how the proposed method can be applied to the wide-breadth of robotic applications that require visual feedback, such as camera-to-robot calibration, robotic tool tracking, and end-effector pose estimation.
Benefiting from the capability of building inter-dependencies among channels or spatial locations, attention mechanisms have been extensively studied and broadly used in a variety of computer vision tasks recently. In this paper, we investigate light-weight but effective attention mechanisms and present triplet attention, a novel method for computing attention weights by capturing cross-dimension interaction using a three-branch structure. For an input tensor, triplet attention builds inter-dimensional dependencies by the rotation operation followed by residual transformations and encodes inter-channel and spatial information with negligible computational overhead. Our method is simple as well as efficient and can be easily plugged into classic backbone networks as an add-on module. We demonstrate the effectiveness of our method on various challenging tasks including image classification on ImageNet-1k and object detection on MSCOCO and PASCAL VOC datasets. Furthermore, we provide extensive in-sight into the performance of triplet attention by visually inspecting the GradCAM and GradCAM++ results. The empirical evaluation of our method supports our intuition on the importance of capturing dependencies across dimensions when computing attention weights. Code for this paper can be publicly accessed at https://github.com/LandskapeAI/triplet-attention
Recent methods for learning unsupervised visual representations, dubbed contrastive learning, optimize the noise-contrastive estimation (NCE) bound on mutual information between two views of an image. NCE uses randomly sampled negative examples to normalize the objective. In this paper, we show that choosing difficult negatives, or those more similar to the current instance, can yield stronger representations. To do this, we introduce a family of mutual information estimators that sample negatives conditionally -- in a "ring" around each positive. We prove that these estimators lower-bound mutual information, with higher bias but lower variance than NCE. Experimentally, we find our approach, applied on top of existing models (IR, CMC, and MoCo) improves accuracy by 2-5% points in each case, measured by linear evaluation on four standard image datasets. Moreover, we find continued benefits when transferring features to a variety of new image distributions from the Meta-Dataset collection and to a variety of downstream tasks such as object detection, instance segmentation, and keypoint detection.
In this paper, we introduce a contrastive learning framework for keypoint detection (CoKe). Keypoint detection differs from other visual tasks where contrastive learning has been applied because the input is a set of images in which multiple keypoints are annotated. This requires the contrastive learning to be extended such that the keypoints are represented and detected independently, which enables the contrastive loss to make the keypoint features different from each other and from the background. Our approach has two benefits: It enables us to exploit contrastive learning for keypoint detection, and by detecting each keypoint independently the detection becomes more robust to occlusion compared to holistic methods, such as stacked hourglass networks, which attempt to detect all keypoints jointly. Our CoKe framework introduces several technical innovations. In particular, we introduce: (i) A clutter bank to represent non-keypoint features; (ii) a keypoint bank that stores prototypical representations of keypoints to approximate the contrastive loss between keypoints; and (iii) a cumulative moving average update to learn the keypoint prototypes while training the feature extractor. Our experiments on a range of diverse datasets (PASCAL3D+, MPII, ObjectNet3D) show that our approach works as well, or better than, alternative methods for keypoint detection, even for human keypoints, for which the literature is vast. Moreover, we observe that CoKe is exceptionally robust to partial occlusion and previously unseen object poses.
This paper addresses the problem of monocular 3D human shape and pose estimation from an RGB image. Despite great progress in this field in terms of pose prediction accuracy, state-of-the-art methods often predict inaccurate body shapes. We suggest that this is primarily due to the scarcity of in-the-wild training data with diverse and accurate body shape labels. Thus, we propose STRAPS (Synthetic Training for Real Accurate Pose and Shape), a system that utilises proxy representations, such as silhouettes and 2D joints, as inputs to a shape and pose regression neural network, which is trained with synthetic training data (generated on-the-fly during training using the SMPL statistical body model) to overcome data scarcity. We bridge the gap between synthetic training inputs and noisy real inputs, which are predicted by keypoint detection and segmentation CNNs at test-time, by using data augmentation and corruption during training. In order to evaluate our approach, we curate and provide a challenging evaluation dataset for monocular human shape estimation, Sports Shape and Pose 3D (SSP-3D). It consists of RGB images of tightly-clothed sports-persons with a variety of body shapes and corresponding pseudo-ground-truth SMPL shape and pose parameters, obtained via multi-frame optimisation. We show that STRAPS outperforms other state-of-the-art methods on SSP-3D in terms of shape prediction accuracy, while remaining competitive with the state-of-the-art on pose-centric datasets and metrics.
Despite the remarkable successes of Convolutional Neural Networks (CNNs) in computer vision, it is time-consuming and error-prone to manually design a CNN. Among various Neural Architecture Search (NAS) methods that are motivated to automate designs of high-performance CNNs, the differentiable NAS and population-based NAS are attracting increasing interests due to their unique characters. To benefit from the merits while overcoming the deficiencies of both, this work proposes a novel NAS method, RelativeNAS. As the key to efficient search, RelativeNAS performs joint learning between fast-learners (i.e. networks with relatively higher accuracy) and slow-learners in a pairwise manner. Moreover, since RelativeNAS only requires low-fidelity performance estimation to distinguish each pair of fast-learner and slow-learner, it saves certain computation costs for training the candidate architectures. The proposed RelativeNAS brings several unique advantages: (1) it achieves state-of-the-art performance on ImageNet with top-1 error rate of 24.88%, i.e. outperforming DARTS and AmoebaNet-B by 1.82% and 1.12% respectively; (2) it spends only nine hours with a single 1080Ti GPU to obtain the discovered cells, i.e. 3.75x and 7875x faster than DARTS and AmoebaNet respectively; (3) it provides that the discovered cells obtained on CIFAR-10 can be directly transferred to object detection, semantic segmentation, and keypoint detection, yielding competitive results of 73.1% mAP on PASCAL VOC, 78.7% mIoU on Cityscapes, and 68.5% AP on MSCOCO, respectively. The implementation of RelativeNAS is available at https://github.com/EMI-Group/RelativeNAS
Have you ever looked at an Instagram model, or a model in a fashion e-commerce web-page, and thought \textit{"Wish I could get a list of fashion items similar to the ones worn by the model!"}. This is what we address in this paper, where we propose a novel computer vision based technique called \textbf{ShopLook} to address the challenging problem of recommending similar fashion products. The proposed method has been evaluated at Myntra (www.myntra.com), a leading online fashion e-commerce platform. In particular, given a user query and the corresponding Product Display Page (PDP) against the query, the goal of our method is to recommend similar fashion products corresponding to the entire set of fashion articles worn by a model in the PDP full-shot image (the one showing the entire model from head to toe). The novelty and strength of our method lies in its capability to recommend similar articles for all the fashion items worn by the model, in addition to the primary article corresponding to the query. This is not only important to promote cross-sells for boosting revenue, but also for improving customer experience and engagement. In addition, our approach is also capable of recommending similar products for User Generated Content (UGC), eg., fashion article images uploaded by users. Formally, our proposed method consists of the following components (in the same order): i) Human keypoint detection, ii) Pose classification, iii) Article localisation and object detection, along with active learning feedback, and iv) Triplet network based image embedding model.
The task of room layout estimation is to locate the wall-floor, wall-ceiling, and wall-wall boundaries. Most recent methods solve this problem based on edge/keypoint detection or semantic segmentation. However, these approaches have shown limited attention on the geometry of the dominant planes and the intersection between them, which has significant impact on room layout. In this work, we propose to incorporate geometric reasoning to deep learning for layout estimation. Our approach learns to infer the depth maps of the dominant planes in the scene by predicting the pixel-level surface parameters, and the layout can be generated by the intersection of the depth maps. Moreover, we present a new dataset with pixel-level depth annotation of dominant planes. It is larger than the existing datasets and contains both cuboid and non-cuboid rooms. Experimental results show that our approach produces considerable performance gains on both 2D and 3D datasets.
Purpose: AI in radiology is hindered chiefly by: 1) Requiring large annotated data sets. 2) Non-generalizability that limits deployment to new scanners / institutions. And 3) Inadequate explainability and interpretability. We believe that reinforcement learning can address all three shortcomings, with robust and intuitive algorithms trainable on small datasets. To the best of our knowledge, reinforcement learning has not been directly applied to computer vision tasks for radiological images. In this proof-of-principle work, we train a deep reinforcement learning network to predict brain tumor location. Materials and Methods: Using the BraTS brain tumor imaging database, we trained a deep Q network on 70 post-contrast T1-weighted 2D image slices. We did so in concert with image exploration, with rewards and punishments designed to localize lesions. To compare with supervised deep learning, we trained a keypoint detection convolutional neural network on the same 70 images. We applied both approaches to a separate 30 image testing set. Results: Reinforcement learning predictions consistently improved during training, whereas those of supervised deep learning quickly diverged. Reinforcement learning predicted testing set lesion locations with 85% accuracy, compared to roughly 7% accuracy for the supervised deep network. Conclusion: Reinforcement learning predicted lesions with high accuracy, which is unprecedented for such a small training set. We believe that reinforcement learning can propel radiology AI well past the inherent limitations of supervised deep learning, with more clinician-driven research and finally toward true clinical applicability.
The attributes of object contours has great significance for instance segmentation task. However, most of the current popular deep neural networks do not pay much attention to the object edge information. Inspired by the human annotation process when making instance segmentation datasets, in this paper, we propose Mask Point R-CNN aiming at promoting the neural network's attention to the object boundary. Specifically, we innovatively extend the original human keypoint detection task to the contour point detection of any object. Based on this analogy, we present an contour point detection auxiliary task to Mask R-CNN, which can boost the gradient flow between different tasks by effectively using feature fusion strategies and multi-task joint training. As a consequence, the model will be more sensitive to the edges of the object and can capture more geometric features. Quantitatively, the experimental results show that our approach outperforms vanilla Mask R-CNN by 3.8\% on Cityscapes dataset and 0.8\% on COCO dataset.
We present a novel end-to-end framework named as GSNet (Geometric and Scene-aware Network), which jointly estimates 6DoF poses and reconstructs detailed 3D car shapes from single urban street view. GSNet utilizes a unique four-way feature extraction and fusion scheme and directly regresses 6DoF poses and shapes in a single forward pass. Extensive experiments show that our diverse feature extraction and fusion scheme can greatly improve model performance. Based on a divide-and-conquer 3D shape representation strategy, GSNet reconstructs 3D vehicle shape with great detail (1352 vertices and 2700 faces). This dense mesh representation further leads us to consider geometrical consistency and scene context, and inspires a new multi-objective loss function to regularize network training, which in turn improves the accuracy of 6D pose estimation and validates the merit of jointly performing both tasks. We evaluate GSNet on the largest multi-task ApolloCar3D benchmark and achieve state-of-the-art performance both quantitatively and qualitatively. Project page is available at https://lkeab.github.io/gsnet/.
Multi-person pose estimation is challenging because it localizes body keypoints for multiple persons simultaneously. Previous methods can be divided into two streams, i.e. top-down and bottom-up methods. The top-down methods localize keypoints after human detection, while the bottom-up methods localize keypoints directly and then cluster/group them for different persons, which are generally more efficient than top-down methods. However, in existing bottom-up methods, the keypoint grouping is usually solved independently from keypoint detection, making them not end-to-end trainable and have sub-optimal performance. In this paper, we investigate a new perspective of human part grouping and reformulate it as a graph clustering task. Especially, we propose a novel differentiable Hierarchical Graph Grouping (HGG) method to learn the graph grouping in bottom-up multi-person pose estimation task. Moreover, HGG is easily embedded into main-stream bottom-up methods. It takes human keypoint candidates as graph nodes and clusters keypoints in a multi-layer graph neural network model. The modules of HGG can be trained end-to-end with the keypoint detection network and is able to supervise the grouping process in a hierarchical manner. To improve the discrimination of the clustering, we add a set of edge discriminators and macro-node discriminators. Extensive experiments on both COCO and OCHuman datasets demonstrate that the proposed method improves the performance of bottom-up pose estimation methods.
The typical bottom-up human pose estimation framework includes two stages, keypoint detection and grouping. Most existing works focus on developing grouping algorithms, e.g., associative embedding, and pixel-wise keypoint regression that we adopt in our approach. We present several schemes that are rarely or unthoroughly studied before for improving keypoint detection and grouping (keypoint regression) performance. First, we exploit the keypoint heatmaps for pixel-wise keypoint regression instead of separating them for improving keypoint regression. Second, we adopt a pixel-wise spatial transformer network to learn adaptive representations for handling the scale and orientation variance to further improve keypoint regression quality. Last, we present a joint shape and heatvalue scoring scheme to promote the estimated poses that are more likely to be true poses. Together with the tradeoff heatmap estimation loss for balancing the background and keypoint pixels and thus improving heatmap estimation quality, we get the state-of-the-art bottom-up human pose estimation result. Code is available at https://github.com/HRNet/HRNet-Bottom-up-Pose-Estimation.
Researchers have attempted utilizing deep neural network (DNN) to learn novel local features from images inspired by its recent successes on a variety of vision tasks. However, existing DNN-based algorithms have not achieved such remarkable progress that could be partly attributed to insufficient utilization of the interactive characters between local feature detector and descriptor. To alleviate these difficulties, we emphasize two desired properties, i.e., repeatability and reliability, to simultaneously summarize the inherent and interactive characters of local feature detector and descriptor. Guided by these properties, a self-supervised framework, namely self-evolving keypoint detection and description (SEKD), is proposed to learn an advanced local feature model from unlabeled natural images. Additionally, to have performance guarantees, novel training strategies have also been dedicatedly designed to minimize the gap between the learned feature and its properties. We benchmark the proposed method on homography estimation, relative pose estimation, and structure-from-motion tasks. Extensive experimental results demonstrate that the proposed method outperforms popular hand-crafted and DNN-based methods by remarkable margins. Ablation studies also verify the effectiveness of each critical training strategy. We will release our code along with the trained model publicly.
This paper introduces a new algorithm for unsupervised learning of keypoint detectors and descriptors, which demonstrates fast convergence and good performance across different datasets. The training procedure uses homographic transformation of images. The proposed model learns to detect points and generate descriptors on pairs of transformed images, which are easy for it to distinguish and repeatedly detect. The trained model follows SuperPoint architecture for ease of comparison, and demonstrates similar performance on natural images from HPatches dataset, and better performance on retina images from Fundus Image Registration Dataset, which contain low number of corner-like features. For HPatches and other datasets, coverage was also computed to provide better estimation of model quality.
In recent years, several unsupervised, "contrastive" learning algorithms in vision have been shown to learn representations that perform remarkably well on transfer tasks. We show that this family of algorithms maximizes a lower bound on the mutual information between two or more "views" of an image where typical views come from a composition of image augmentations. Our bound generalizes the InfoNCE objective to support negative sampling from a restricted region of "difficult" contrasts. We find that the choice of negative samples and views are critical to the success of these algorithms. Reformulating previous learning objectives in terms of mutual information also simplifies and stabilizes them. In practice, our new objectives yield representations that outperform those learned with previous approaches for transfer to classification, bounding box detection, instance segmentation, and keypoint detection. % experiments show that choosing more difficult negative samples results in a stronger representation, outperforming those learned with IR, LA, and CMC in classification, bounding box detection, instance segmentation, and keypoint detection. The mutual information framework provides a unifying comparison of approaches to contrastive learning and uncovers the choices that impact representation learning.
Keypoint detection and description is fundamental yet important in many vision applications. Most existing methods use detect-then-describe or detect-and-describe strategy to learn local features without considering their context information. Consequently, it is challenging for these methods to learn robust local features. In this paper, we focus on the fusion of low-level textual information and high-level semantic context information to improve the discrimitiveness of local features. Specifically, we first estimate a score map to represent the distribution of potential keypoints according to the quality of descriptors of all pixels. Then, we extract and aggregate multi-scale high-level semantic features based by the guidance of the score map. Finally, the low-level local features and high-level semantic features are fused and refined using a residual module. Experiments on the challenging local feature benchmark dataset demonstrate that our method achieves the state-of-the-art performance in the local feature challenge of the visual localization benchmark.
Understanding interactions between humans and objects is one of the fundamental problems in visual classification and an essential step towards detailed scene understanding. Human-object interaction (HOI) detection strives to localize both the human and an object as well as the identification of complex interactions between them. Most existing HOI detection approaches are instance-centric where interactions between all possible human-object pairs are predicted based on appearance features and coarse spatial information. We argue that appearance features alone are insufficient to capture complex human-object interactions. In this paper, we therefore propose a novel fully-convolutional approach that directly detects the interactions between human-object pairs. Our network predicts interaction points, which directly localize and classify the inter-action. Paired with the densely predicted interaction vectors, the interactions are associated with human and object detections to obtain final predictions. To the best of our knowledge, we are the first to propose an approach where HOI detection is posed as a keypoint detection and grouping problem. Experiments are performed on two popular benchmarks: V-COCO and HICO-DET. Our approach sets a new state-of-the-art on both datasets. Code is available at https://github.com/vaesl/IP-Net.