While short range 3D pedestrian detection is sufficient for emergency breaking, long range detections are required for smooth breaking and gaining trust in autonomous vehicles. The current state-of-the-art on the KITTI benchmark performs suboptimal in detecting the position of pedestrians at long range. Thus, we propose an approach specifically targeting long range 3D pedestrian detection (LRPD), leveraging the density of RGB and the precision of LiDAR. Therefore, for proposals, RGB instance segmentation and LiDAR point based proposal generation are combined, followed by a second stage using both sensor modalities symmetrically. This leads to a significant improvement in mAP on long range compared to the current state-of-the art. The evaluation of our LRPD approach was done on the pedestrians from the KITTI benchmark.
We propose a new regularization method to alleviate over-fitting in deep neural networks. The key idea is utilizing randomly transformed training samples to regularize a set of sub-networks, which are originated by sampling the width of the original network, in the training process. As such, the proposed method introduces self-guided disturbances to the raw gradients of the network and therefore is termed as Gradient Augmentation (GradAug). We demonstrate that GradAug can help the network learn well-generalized and more diverse representations. Moreover, it is easy to implement and can be applied to various structures and applications. GradAug improves ResNet-50 to 78.79% on ImageNet classification, which is a new state-of-the-art accuracy. By combining with CutMix, it further boosts the performance to 79.67%, which outperforms an ensemble of advanced training tricks. The generalization ability is evaluated on COCO object detection and instance segmentation where GradAug significantly surpasses other state-of-the-art methods. GradAug is also robust to image distortions and FGSM adversarial attacks and is highly effective in low data regimes. Code is available at https://github.com/taoyang1122/GradAug
Most previous works of outdoor instance segmentation for images only use color information. We explore a novel direction of sensor fusion to exploit stereo cameras. Geometric information from disparities helps separate overlapping objects of the same or different classes. Moreover, geometric information penalizes region proposals with unlikely 3D shapes thus suppressing false positive detections. Mask regression is based on 2D, 2.5D, and 3D ROI using the pseudo-lidar and image-based representations. These mask predictions are fused by a mask scoring process. However, public datasets only adopt stereo systems with shorter baseline and focal legnth, which limit measuring ranges of stereo cameras. We collect and utilize High-Quality Driving Stereo (HQDS) dataset, using much longer baseline and focal length with higher resolution. Our performance attains state of the art. Please refer to our project page. The full paper is available here.
In this work, we address the problem of multi-person 3D pose estimation from a single image. A typical regression approach in the top-down setting of this problem would first detect all humans and then reconstruct each one of them independently. However, this type of prediction suffers from incoherent results, e.g., interpenetration and inconsistent depth ordering between the people in the scene. Our goal is to train a single network that learns to avoid these problems and generate a coherent 3D reconstruction of all the humans in the scene. To this end, a key design choice is the incorporation of the SMPL parametric body model in our top-down framework, which enables the use of two novel losses. First, a distance field-based collision loss penalizes interpenetration among the reconstructed people. Second, a depth ordering-aware loss reasons about occlusions and promotes a depth ordering of people that leads to a rendering which is consistent with the annotated instance segmentation. This provides depth supervision signals to the network, even if the image has no explicit 3D annotations. The experiments show that our approach outperforms previous methods on standard 3D pose benchmarks, while our proposed losses enable more coherent reconstruction in natural images. The project website with videos, results, and code can be found at: https://jiangwenpl.github.io/multiperson
The de-facto approach to many vision tasks is to start from pretrained visual representations, typically learned via supervised training on ImageNet. Recent methods have explored unsupervised pretraining to scale to vast quantities of unlabeled images. In contrast, we aim to learn high-quality visual representations from fewer images. To this end, we revisit supervised pretraining, and seek data-efficient alternatives to classification-based pretraining. We propose VirTex -- a pretraining approach using semantically dense captions to learn visual representations. We train convolutional networks from scratch on COCO Captions, and transfer them to downstream recognition tasks including image classification, object detection, and instance segmentation. On all tasks, VirTex yields features that match or exceed those learned on ImageNet -- supervised or unsupervised -- despite using up to ten times fewer images.
Building segmentation is of great importance in the task of remote sensing imagery interpretation. However, the existing semantic segmentation and instance segmentation methods often lead to segmentation masks with blurred boundaries. In this paper, we propose a novel instance segmentation network for building segmentation in high-resolution remote sensing images. More specifically, we consider segmenting an individual building as detecting several keypoints. The detected keypoints are subsequently reformulated as a closed polygon, which is the semantic boundary of the building. By doing so, the sharp boundary of the building could be preserved. Experiments are conducted on selected Aerial Imagery for Roof Segmentation (AIRS) dataset, and our method achieves better performance in both quantitative and qualitative results with comparison to the state-of-the-art methods. Our network is a bottom-up instance segmentation method that could well preserve geometric details.
Nucleus segmentation is an important task in medical image analysis. However, machine learning models cannot perform well because there are large amount of clusters of crowded nuclei. To handle this problem, existing approaches typically resort to sophisticated hand-crafted post-processing strategies; therefore, they are vulnerable to the variation of post-processing hyper-parameters. Accordingly, in this paper, we devise a Boundary-assisted Region Proposal Network (BRP-Net) that achieves robust instance-level nucleus segmentation. First, we propose a novel Task-aware Feature Encoding (TAFE) network that efficiently extracts respective high-quality features for semantic segmentation and instance boundary detection tasks. This is achieved by carefully considering the correlation and differences between the two tasks. Second, coarse nucleus proposals are generated based on the predictions of the above two tasks. Third, these proposals are fed into instance segmentation networks for more accurate prediction. Experimental results demonstrate that the performance of BRP-Net is robust to the variation of post-processing hyper-parameters. Furthermore, BRP-Net achieves state-of-the-art performances on both the Kumar and CPM17 datasets. The code of BRP-Net will be released at https://github.com/csccsccsccsc/brpnet.
Many modern object detectors demonstrate outstanding performances by using the mechanism of looking and thinking twice. In this paper, we explore this mechanism in the backbone design for object detection. At the macro level, we propose Recursive Feature Pyramid, which incorporates extra feedback connections from Feature Pyramid Networks into the bottom-up backbone layers. At the micro level, we propose Switchable Atrous Convolution, which convolves the features with different atrous rates and gathers the results using switch functions. Combining them results in DetectoRS, which significantly improves the performances of object detection. On COCO test-dev, DetectoRS achieves state-of-the-art 55.7% box AP for object detection, 48.5% mask AP for instance segmentation, and 50.0% PQ for panoptic segmentation. The code is made publicly available.
A major challenge in clinical In-Vitro Fertilization (IVF) is selecting the highest quality embryo to transfer to the patient in the hopes of achieving a pregnancy. Time-lapse microscopy provides clinicians with a wealth of information for selecting embryos. However, the resulting movies of embryos are currently analyzed manually, which is time consuming and subjective. Here, we automate feature extraction of time-lapse microscopy of human embryos with a machine-learning pipeline of five convolutional neural networks (CNNs). Our pipeline consists of (1) semantic segmentation of the regions of the embryo, (2) regression predictions of fragment severity, (3) classification of the developmental stage, and object instance segmentation of (4) cells and (5) pronuclei. Our approach greatly speeds up the measurement of quantitative, biologically relevant features that may aid in embryo selection.
Object segmentation is an important step in the workflow of computational pathology. Deep learning based models generally require large amount of labeled data for precise and reliable prediction. However, collecting labeled data is expensive because it often requires expert knowledge, particularly in medical imaging domain where labels are the result of a time-consuming analysis made by one or more human experts. As nuclei, cells and glands are fundamental objects for downstream analysis in computational pathology/cytology, in this paper we propose a simple CNN-based approach to speed up collecting annotations for these objects which requires minimum interaction from the annotator. We show that for nuclei and cells in histology and cytology images, one click inside each object is enough for NuClick to yield a precise annotation. For multicellular structures such as glands, we propose a novel approach to provide the NuClick with a squiggle as a guiding signal, enabling it to segment the glandular boundaries. These supervisory signals are fed to the network as auxiliary inputs along with RGB channels. With detailed experiments, we show that NuClick is adaptable to the object scale, robust against variations in the user input, adaptable to new domains, and delivers reliable annotations. An instance segmentation model trained on masks generated by NuClick achieved the first rank in LYON19 challenge. As exemplar outputs of our framework, we are releasing two datasets: 1) a dataset of lymphocyte annotations within IHC images, and 2) a dataset of segmented WBCs in blood smear images.
We introduce the Northumberland Dolphin Dataset 2020 (NDD20), a challenging image dataset annotated for both coarse and fine-grained instance segmentation and categorisation. This dataset, the first release of the NDD, was created in response to the rapid expansion of computer vision into conservation research and the production of field-deployable systems suited to extreme environmental conditions -- an area with few open source datasets. NDD20 contains a large collection of above and below water images of two different dolphin species for traditional coarse and fine-grained segmentation. All data contained in NDD20 was obtained via manual collection in the North Sea around the Northumberland coastline, UK. We present experimentation using standard deep learning network architecture trained using NDD20 and report baselines results.
This paper presents a fully automatic framework for extracting editable 3D objects directly from a single photograph. Unlike previous methods which recover either depth maps, point clouds, or mesh surfaces, we aim to recover 3D objects with semantic parts and can be directly edited. We base our work on the assumption that most human-made objects are constituted by parts and these parts can be well represented by generalized primitives. Our work makes an attempt towards recovering two types of primitive-shaped objects, namely, generalized cuboids and generalized cylinders. To this end, we build a novel instance-aware segmentation network for accurate part separation. Our GeoNet outputs a set of smooth part-level masks labeled as profiles and bodies. Then in a key stage, we simultaneously identify profile-body relations and recover 3D parts by sweeping the recognized profile along their body contour and jointly optimize the geometry to align with the recovered masks. Qualitative and quantitative experiments show that our algorithm can recover high quality 3D models and outperforms existing methods in both instance segmentation and 3D reconstruction. The dataset and code of AutoSweep are available at https://chenxin.tech/AutoSweep.html.
We present a new version of YOLO with better performance and extended with instance segmentation called Poly-YOLO. Poly-YOLO builds on the original ideas of YOLOv3 and removes two of its weaknesses: a large amount of rewritten labels and inefficient distribution of anchors. Poly-YOLO reduces the issues by aggregating features from a light SE-Darknet-53 backbone with a hypercolumn technique, using stairstep upsampling, and produces a single scale output with high resolution. In comparison with YOLOv3, Poly-YOLO has only 60% of its trainable parameters but improves mAP by a relative 40%. We also present Poly-YOLO lite with fewer parameters and a lower output resolution. It has the same precision as YOLOv3, but it is three times smaller and twice as fast, thus suitable for embedded devices. Finally, Poly-YOLO performs instance segmentation using bounding polygons. The network is trained to detect size-independent polygons defined on a polar grid. Vertices of each polygon are being predicted with their confidence, and therefore Poly-YOLO produces polygons with a varying number of vertices.
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.
For object detection, how to address the contradictory requirement between feature map resolution and receptive field on high-resolution inputs still remains an open question. In this paper, to tackle this issue, we build a novel architecture, called Attention-guided Context Feature Pyramid Network (AC-FPN), that exploits discriminative information from various large receptive fields via integrating attention-guided multi-path features. The model contains two modules. The first one is Context Extraction Module (CEM) that explores large contextual information from multiple receptive fields. As redundant contextual relations may mislead localization and recognition, we also design the second module named Attention-guided Module (AM), which can adaptively capture the salient dependencies over objects by using the attention mechanism. AM consists of two sub-modules, i.e., Context Attention Module (CxAM) and Content Attention Module (CnAM), which focus on capturing discriminative semantics and locating precise positions, respectively. Most importantly, our AC-FPN can be readily plugged into existing FPN-based models. Extensive experiments on object detection and instance segmentation show that existing models with our proposed CEM and AM significantly surpass their counterparts without them, and our model successfully obtains state-of-the-art results. We have released the source code at https://github.com/Caojunxu/AC-FPN.
Histo-pathological diagnostics are an inherent part of the everyday work but are particularly laborious and associated with time-consuming manual analysis of image data. In order to cope with the increasing diagnostic case numbers due to the current growth and demographic change of the global population and the progress in personalized medicine, pathologists ask for assistance. Profiting from digital pathology and the use of artificial intelligence, individual solutions can be offered (e.g. detect labeled cancer tissue sections). The testing of the human epidermal growth factor receptor 2 (HER2) oncogene amplification status via fluorescence in situ hybridization (FISH) is recommended for breast and gastric cancer diagnostics and is regularly performed at clinics. Here, we develop an interpretable, deep learning (DL)-based pipeline which automates the evaluation of FISH images with respect to HER2 gene amplification testing. It mimics the pathological assessment and relies on the detection and localization of interphase nuclei based on instance segmentation networks. Furthermore, it localizes and classifies fluorescence signals within each nucleus with the help of image classification and object detection convolutional neural networks (CNNs). Finally, the pipeline classifies the whole image regarding its HER2 amplification status. The visualization of pixels on which the networks' decision occurs, complements an essential part to enable interpretability by pathologists.
The behavioural research of pigs can be greatly simplified if automatic recognition systems are used. Especially systems based on computer vision have the advantage that they allow an evaluation without affecting the normal behaviour of the animals. In recent years, methods based on deep learning have been introduced and have shown pleasingly good results. Especially object and keypoint detectors have been used to detect the individual animals. Despite good results, bounding boxes and sparse keypoints do not trace the contours of the animals, resulting in a lot of information being lost. Therefore this work follows the relatively new definition of a panoptic segmentation and aims at the pixel accurate segmentation of the individual pigs. For this a framework of a neural network for semantic segmentation, different network heads and postprocessing methods is presented. With the resulting instance segmentation masks further information like the size or weight of the animals could be estimated. The method is tested on a specially created data set with 1000 hand-labeled images and achieves detection rates of around 95% (F1 Score) despite disturbances such as occlusions and dirty lenses.
Embodied perception refers to the ability of an autonomous agent to perceive its environment so that it can (re)act. The responsiveness of the agent is largely governed by latency of its processing pipeline. While past work has studied the algorithmic trade-off between latency and accuracy, there has not been a clear metric to compare different methods along the Pareto optimal latency-accuracy curve. We point out a discrepancy between standard offline evaluation and real-time applications: by the time an algorithm finishes processing a particular frame, the surrounding world has changed. To these ends, we present an approach that coherently integrates latency and accuracy into a single metric for real-time online perception, which we refer to as "streaming accuracy". The key insight behind this metric is to jointly evaluate the output of the entire perception stack at every time instant, forcing the stack to consider the amount of streaming data that should be ignored while computation is occurring. More broadly, building upon this metric, we introduce a meta-benchmark that systematically converts any single-frame task into a streaming perception task. We focus on the illustrative tasks of object detection and instance segmentation in urban video streams, and contribute a novel dataset with high-quality and temporally-dense annotations. Our proposed solutions and their empirical analysis demonstrate a number of surprising conclusions: (1) there exists an optimal "sweet spot" that maximizes streaming accuracy along the Pareto optimal latency-accuracy curve, (2) asynchronous tracking and future forecasting naturally emerge as internal representations that enable streaming perception, and (3) dynamic scheduling can be used to overcome temporal aliasing, yielding the paradoxical result that latency is sometimes minimized by sitting idle and "doing nothing".
Contrastive learning between multiple views of the data has recently achieved state of the art performance in the field of self-supervised representation learning. Despite its success, the influence of different view choices has been less studied. In this paper, we use theoretical and empirical analysis to better understand the importance of view selection, and argue that we should reduce the mutual information (MI) between views while keeping task-relevant information intact. To verify this hypothesis, we devise unsupervised and semi-supervised frameworks that learn effective views by aiming to reduce their MI. We also consider data augmentation as a way to reduce MI, and show that increasing data augmentation indeed leads to decreasing MI and improves downstream classification accuracy. As a by-product, we achieve a new state-of-the-art accuracy on unsupervised pre-training for ImageNet classification ($73\%$ top-1 linear readout with a ResNet-50). In addition, transferring our models to PASCAL VOC object detection and COCO instance segmentation consistently outperforms supervised pre-training. Code:http://github.com/HobbitLong/PyContrast
Instance segmentation of unknown objects from images is regarded as relevant for several robot skills including grasping, tracking and object sorting. Recent results in computer vision have shown that large hand-labeled datasets enable high segmentation performance. To overcome the time-consuming process of manually labeling data for new environments, we present a transfer learning approach for robots that learn to segment objects by interacting with their environment in a self-supervised manner. Our robot pushes unknown objects on a table and uses information from optical flow to create training labels in the form of object masks. To achieve this, we fine-tune an existing DeepMask network for instance segmentation on the self-labeled training data acquired by the robot. We evaluate our trained network (SelfDeepMask) on a set of real images showing challenging and cluttered scenes with novel objects. Here, SelfDeepMask outperforms the DeepMask network trained on the COCO dataset by 9.5% in average precision. Furthermore, we combine our approach with recent approaches for training with noisy labels in order to better cope with induced label noise.