Detecting tiny objects is a very challenging problem since a tiny object only contains a few pixels in size. We demonstrate that state-of-the-art detectors do not produce satisfactory results on tiny objects due to the lack of appearance information. Our key observation is that Intersection over Union (IoU) based metrics such as IoU itself and its extensions are very sensitive to the location deviation of the tiny objects, and drastically deteriorate the detection performance when used in anchor-based detectors. To alleviate this, we propose a new evaluation metric using Wasserstein distance for tiny object detection. Specifically, we first model the bounding boxes as 2D Gaussian distributions and then propose a new metric dubbed Normalized Wasserstein Distance (NWD) to compute the similarity between them by their corresponding Gaussian distributions. The proposed NWD metric can be easily embedded into the assignment, non-maximum suppression, and loss function of any anchor-based detector to replace the commonly used IoU metric. We evaluate our metric on a new dataset for tiny object detection (AI-TOD) in which the average object size is much smaller than existing object detection datasets. Extensive experiments show that, when equipped with NWD metric, our approach yields performance that is 6.7 AP points higher than a standard fine-tuning baseline, and 6.0 AP points higher than state-of-the-art competitors. Codes are available at: https://github.com/jwwangchn/NWD.
Aiming at recognizing and localizing the object of novel categories by a few reference samples, few-shot object detection (FSOD) is a quite challenging task. Previous works often depend on the fine-tuning process to transfer their model to the novel category and rarely consider the defect of fine-tuning, resulting in many application drawbacks. For example, these methods are far from satisfying in the episode-changeable scenarios due to excessive fine-tuning times, and their performance on low-quality (e.g., low-shot and class-incomplete) support sets degrades severely. To this end, this paper proposes an instant response few-shot object detector (IR-FSOD) that can accurately and directly detect the objects of novel categories without the fine-tuning process. To accomplish the objective, we carefully analyze the defects of individual modules in the Faster R-CNN framework under the FSOD setting and then extend it to IR-FSOD by improving these defects. Specifically, we first propose two simple but effective meta-strategies for the box classifier and RPN module to enable the object detection of novel categories with instant response. Then, we introduce two explicit inferences into the localization module to alleviate its over-fitting to the base categories, including explicit localization score and semi-explicit box regression. Extensive experiments show that the IR-FSOD framework not only achieves few-shot object detection with the instant response but also reaches state-of-the-art performance in precision and recall under various FSOD settings.
Incorporating relational reasoning in neural networks for object recognition remains an open problem. Although many attempts have been made for relational reasoning, they generally only consider a single type of relationship. For example, pixel relations through self-attention (e.g., non-local networks), scale relations through feature fusion (e.g., feature pyramid networks), or object relations through graph convolutions (e.g., reasoning-RCNN). Little attention has been given to more generalized frameworks that can reason across these relationships. In this paper, we propose a hierarchical relational reasoning framework (HR-RCNN) for object detection, which utilizes a novel graph attention module (GAM). This GAM is a concise module that enables reasoning across heterogeneous nodes by operating on the graph edges directly. Leveraging heterogeneous relationships, our HR-RCNN shows great improvement on COCO dataset, for both object detection and instance segmentation.
Knowledge distillation has shown great success in classification, however, it is still challenging for detection. In a typical image for detection, representations from different locations may have different contributions to detection targets, making the distillation hard to balance. In this paper, we propose a conditional distillation framework to distill the desired knowledge, namely knowledge that is beneficial in terms of both classification and localization for every instance. The framework introduces a learnable conditional decoding module, which retrieves information given each target instance as query. Specifically, we encode the condition information as query and use the teacher's representations as key. The attention between query and key is used to measure the contribution of different features, guided by a localization-recognition-sensitive auxiliary task. Extensive experiments demonstrate the efficacy of our method: we observe impressive improvements under various settings. Notably, we boost RetinaNet with ResNet-50 backbone from 37.4 to 40.7 mAP (+3.3) under 1x schedule, that even surpasses the teacher (40.4 mAP) with ResNet-101 backbone under 3x schedule. Code has been released on https://github.com/megvii-research/ICD.
Owing to effective and flexible data acquisition, unmanned aerial vehicle (UAV) has recently become a hotspot across the fields of computer vision (CV) and remote sensing (RS). Inspired by recent success of deep learning (DL), many advanced object detection and tracking approaches have been widely applied to various UAV-related tasks, such as environmental monitoring, precision agriculture, traffic management. This paper provides a comprehensive survey on the research progress and prospects of DL-based UAV object detection and tracking methods. More specifically, we first outline the challenges, statistics of existing methods, and provide solutions from the perspectives of DL-based models in three research topics: object detection from the image, object detection from the video, and object tracking from the video. Open datasets related to UAV-dominated object detection and tracking are exhausted, and four benchmark datasets are employed for performance evaluation using some state-of-the-art methods. Finally, prospects and considerations for the future work are discussed and summarized. It is expected that this survey can facilitate those researchers who come from remote sensing field with an overview of DL-based UAV object detection and tracking methods, along with some thoughts on their further developments.
3D image processing is an important problem in computer vision and pattern recognition fields. Compared with 2D image processing, its computation difficulty and cost are much higher due to the extra dimension. To fundamentally address this problem, we propose to embed the essential information in a 3D object into 2D space via spectral layout. Specifically, we construct a 3D adjacency graph to capture spatial structure of the 3D voxel grid. Then we calculate the eigenvectors corresponding to the second and third smallest eigenvalues of its graph Laplacian and perform spectral layout to map each voxel into a pixel in 2D Cartesian coordinate plane. The proposed method can achieve high quality 2D representations for 3D objects, which enables to use 2D-based methods to process 3D objects. The experimental results demonstrate the effectiveness and efficiency of our method.
Neural Architecture Search (NAS) has shown great potential in effectively reducing manual effort in network design by automatically discovering optimal architectures. What is noteworthy is that as of now, object detection is less touched by NAS algorithms despite its significant importance in computer vision. To the best of our knowledge, most of the recent NAS studies on object detection tasks fail to satisfactorily strike a balance between performance and efficiency of the resulting models, let alone the excessive amount of computational resources cost by those algorithms. Here we propose an efficient method to obtain better object detectors by searching for the feature pyramid network (FPN) as well as the prediction head of a simple anchor-free object detector, namely, FCOS [36], using a tailored reinforcement learning paradigm. With carefully designed search space, search algorithms, and strategies for evaluating network quality, we are able to find top-performing detection architectures within 4 days using 8 V100 GPUs. The discovered architectures surpass state-of-the-art object detection models (such as Faster R-CNN, Retina-Net and, FCOS) by 1.0% to 5.4% points in AP on the COCO dataset, with comparable computation complexity and memory footprint, demonstrating the efficacy of the proposed NAS method for object detection. Code is available at https://github.com/Lausannen/NAS-FCOS.
Deep Neural Networks (DNNs) have been shown to be vulnerable to adversarial examples, which are slightly perturbed input images which lead DNNs to make wrong predictions. To protect from such examples, various defense strategies have been proposed. A very recent defense strategy for detecting adversarial examples, that has been shown to be robust to current attacks, is to check for intrinsic context consistencies in the input data, where context refers to various relationships (e.g., object-to-object co-occurrence relationships) in images. In this paper, we show that even context consistency checks can be brittle to properly crafted adversarial examples and to the best of our knowledge, we are the first to do so. Specifically, we propose an adaptive framework to generate examples that subvert such defenses, namely, Adversarial attacks against object Detection that evade Context consistency checks (ADC). In ADC, we formulate a joint optimization problem which has two attack goals, viz., (i) fooling the object detector and (ii) evading the context consistency check system, at the same time. Experiments on both PASCAL VOC and MS COCO datasets show that examples generated with ADC fool the object detector with a success rate of over 85% in most cases, and at the same time evade the recently proposed context consistency checks, with a bypassing rate of over 80% in most cases. Our results suggest that how to robustly model context and check its consistency, is still an open problem.
We present an improvement in the ability to meaningfully track features in high speed videos of Caribbean steelpan drums illuminated by Electronic Speckle Pattern Interferometry (ESPI). This is achieved through the use of up-to-date computer vision libraries for object detection and image segmentation as well as a significant effort toward cleaning the dataset previously used to train systems for this application. Besides improvements on previous metric scores by 10% or more, noteworthy in this project are the introduction of a segmentation-regression map for the entire drum surface yielding interference fringe counts comparable to those obtained via object detection, as well as the accelerated workflow for coordinating the data-cleaning-and-model-training feedback loop for rapid iteration allowing this project to be conducted on a timescale of only 18 days.
Video relation detection forms a new and challenging problem in computer vision, where subjects and objects need to be localized spatio-temporally and a predicate label needs to be assigned if and only if there is an interaction between the two. Despite recent progress in video relation detection, overall performance is still marginal and it remains unclear what the key factors are towards solving the problem. Following examples set in the object detection and action localization literature, we perform a deep dive into the error diagnosis of current video relation detection approaches. We introduce a diagnostic tool for analyzing the sources of detection errors. Our tool evaluates and compares current approaches beyond the single scalar metric of mean Average Precision by defining different error types specific to video relation detection, used for false positive analyses. Moreover, we examine different factors of influence on the performance in a false negative analysis, including relation length, number of subject/object/predicate instances, and subject/object size. Finally, we present the effect on video relation performance when considering an oracle fix for each error type. On two video relation benchmarks, we show where current approaches excel and fall short, allowing us to pinpoint the most important future directions in the field. The tool is available at \url{https://github.com/shanshuo/DiagnoseVRD}.
Inspired by the great success achieved by CNN in image recognition, view-based methods applied CNNs to model the projected views for 3D object understanding and achieved excellent performance. Nevertheless, multi-view CNN models cannot model the communications between patches from different views, limiting its effectiveness in 3D object recognition. Inspired by the recent success gained by vision Transformer in image recognition, we propose a Multi-view Vision Transformer (MVT) for 3D object recognition. Since each patch feature in a Transformer block has a global reception field, it naturally achieves communications between patches from different views. Meanwhile, it takes much less inductive bias compared with its CNN counterparts. Considering both effectiveness and efficiency, we develop a global-local structure for our MVT. Our experiments on two public benchmarks, ModelNet40 and ModelNet10, demonstrate the competitive performance of our MVT.
Holistic object representation-based trackers suffer from performance drop under large appearance change such as deformation and occlusion. In this work, we propose a dynamic part-based tracker and constantly update the target part representation to adapt to object appearance change. Moreover, we design an attention-guided part localization network to directly predict the target part locations, and determine the final bounding box with the distribution of target parts. Our proposed tracker achieves promising results on various benchmarks: VOT2018, OTB100 and GOT-10k
Box representation has been extensively used for object detection in computer vision. Such representation is efficacious but not necessarily optimized for biomedical objects (e.g., glomeruli), which play an essential role in renal pathology. In this paper, we propose a simple circle representation for medical object detection and introduce CircleNet, an anchor-free detection framework. Compared with the conventional bounding box representation, the proposed bounding circle representation innovates in three-fold: (1) it is optimized for ball-shaped biomedical objects; (2) The circle representation reduced the degree of freedom compared with box representation; (3) It is naturally more rotation invariant. When detecting glomeruli and nuclei on pathological images, the proposed circle representation achieved superior detection performance and be more rotation-invariant, compared with the bounding box. The code has been made publicly available: https://github.com/hrlblab/CircleNet
Visual Emotion Analysis (VEA) aims at finding out how people feel emotionally towards different visual stimuli, which has attracted great attention recently with the prevalence of sharing images on social networks. Since human emotion involves a highly complex and abstract cognitive process, it is difficult to infer visual emotions directly from holistic or regional features in affective images. It has been demonstrated in psychology that visual emotions are evoked by the interactions between objects as well as the interactions between objects and scenes within an image. Inspired by this, we propose a novel Scene-Object interreLated Visual Emotion Reasoning network (SOLVER) to predict emotions from images. To mine the emotional relationships between distinct objects, we first build up an Emotion Graph based on semantic concepts and visual features. Then, we conduct reasoning on the Emotion Graph using Graph Convolutional Network (GCN), yielding emotion-enhanced object features. We also design a Scene-Object Fusion Module to integrate scenes and objects, which exploits scene features to guide the fusion process of object features with the proposed scene-based attention mechanism. Extensive experiments and comparisons are conducted on eight public visual emotion datasets, and the results demonstrate that the proposed SOLVER consistently outperforms the state-of-the-art methods by a large margin. Ablation studies verify the effectiveness of our method and visualizations prove its interpretability, which also bring new insight to explore the mysteries in VEA. Notably, we further discuss SOLVER on three other potential datasets with extended experiments, where we validate the robustness of our method and notice some limitations of it.
Although DALL-E has shown an impressive ability of composition-based systematic generalization in image generation, it requires the dataset of text-image pairs and the compositionality is provided by the text. In contrast, object-centric representation models like the Slot Attention model learn composable representations without the text prompt. However, unlike DALL-E its ability to systematically generalize for zero-shot generation is significantly limited. In this paper, we propose a simple but novel slot-based autoencoding architecture, called SLATE, for combining the best of both worlds: learning object-centric representations that allows systematic generalization in zero-shot image generation without text. As such, this model can also be seen as an illiterate DALL-E model. Unlike the pixel-mixture decoders of existing object-centric representation models, we propose to use the Image GPT decoder conditioned on the slots for capturing complex interactions among the slots and pixels. In experiments, we show that this simple and easy-to-implement architecture not requiring a text prompt achieves significant improvement in in-distribution and out-of-distribution (zero-shot) image generation and qualitatively comparable or better slot-attention structure than the models based on mixture decoders.
Depth (D) indicates occlusion and is less sensitive to illumination changes, which make depth attractive modality for Visual Object Tracking (VOT). Depth is used in RGBD object tracking where the best trackers are deep RGB trackers with additional heuristic using depth maps. There are two potential reasons for the heuristics: 1) the lack of large RGBD tracking datasets to train deep RGBD trackers and 2) the long-term evaluation protocol of VOT RGBD that benefits from heuristics such as depth-based occlusion detection. In this work, we study how far D-only tracking can go if trained with large amounts of depth data. To compensate the lack of depth data, we generate depth maps for tracking. We train a "Depth-DiMP" from the scratch with the generated data and fine-tune it with the available small RGBD tracking datasets. The depth-only DiMP achieves good accuracy in depth-only tracking and combined with the original RGB DiMP the end-to-end trained RGBD-DiMP outperforms the recent VOT 2020 RGBD winners.
Practical object pose estimation demands robustness against occlusions to the target object. State-of-the-art (SOTA) object pose estimators take a two-stage approach, where the first stage predicts 2D landmarks using a deep network and the second stage solves for 6DOF pose from 2D-3D correspondences. Albeit widely adopted, such two-stage approaches could suffer from novel occlusions when generalising and weak landmark coherence due to disrupted features. To address these issues, we develop a novel occlude-and-blackout batch augmentation technique to learn occlusion-robust deep features, and a multi-precision supervision architecture to encourage holistic pose representation learning for accurate and coherent landmark predictions. We perform careful ablation tests to verify the impact of our innovations and compare our method to SOTA pose estimators. Without the need of any post-processing or refinement, our method exhibits superior performance on the LINEMOD dataset. On the YCB-Video dataset our method outperforms all non-refinement methods in terms of the ADD(-S) metric. We also demonstrate the high data-efficiency of our method. Our code is available at http://github.com/BoChenYS/ROPE
A key component of understanding hand-object interactions is the ability to identify the active object -- the object that is being manipulated by the human hand. In order to accurately localize the active object, any method must reason using information encoded by each image pixel, such as whether it belongs to the hand, the object, or the background. To leverage each pixel as evidence to determine the bounding box of the active object, we propose a pixel-wise voting function. Our pixel-wise voting function takes an initial bounding box as input and produces an improved bounding box of the active object as output. The voting function is designed so that each pixel inside of the input bounding box votes for an improved bounding box, and the box with the majority vote is selected as the output. We call the collection of bounding boxes generated inside of the voting function, the Relational Box Field, as it characterizes a field of bounding boxes defined in relationship to the current bounding box. While our voting function is able to improve the bounding box of the active object, one round of voting is typically not enough to accurately localize the active object. Therefore, we repeatedly apply the voting function to sequentially improve the location of the bounding box. However, since it is known that repeatedly applying a one-step predictor (i.e., auto-regressive processing with our voting function) can cause a data distribution shift, we mitigate this issue using reinforcement learning (RL). We adopt standard RL to learn the voting function parameters and show that it provides a meaningful improvement over a standard supervised learning approach. We perform experiments on two large-scale datasets: 100DOH and MECCANO, improving AP50 performance by 8% and 30%, respectively, over the state of the art.
Object Detection on the mobile system is a challenge in terms of everything. Nowadays, many object detection models have been designed, and most of them concentrate on precision. However, the computation burden of those models on mobile systems is unacceptable. Researchers have designed some lightweight networks for mobiles by sacrificing precision. We present a novel edge-based task scheduling framework for object detection (termed as ESOD). In detail, we train a DNN model (termed as pre-model) to predict which object detection model to use for the coming task and offloads to which edge servers by physical characteristics of the image task (e.g., brightness, saturation). The results show that ESOD can reduce latency and energy consumption by an average of 22.13% and 29.60% and improve the mAP to 45.8(with 0.9 mAP better), respectively, compared with the SOTA DETR model.
Recent work indicates that, besides being a challenge in producing perceptually pleasing images, low light proves more difficult for machine cognition than previously thought. In our work, we take a closer look at object detection in low light. First, to support the development and evaluation of new methods in this domain, we present a high-quality large-scale Night Object Detection (NOD) dataset showing dynamic scenes captured on the streets at night. Next, we directly link the lighting conditions to perceptual difficulty and identify what makes low light problematic for machine cognition. Accordingly, we provide instance-level annotation for a subset of the dataset for an in-depth evaluation of future methods. We also present an analysis of the baseline model performance to highlight opportunities for future research and show that low light is a non-trivial problem that requires special attention from the researchers. Further, to address the issues caused by low light, we propose to incorporate an image enhancement module into the object detection framework and two novel data augmentation techniques. Our image enhancement module is trained under the guidance of the object detector to learn image representation optimal for machine cognition rather than for the human visual system. Finally, experimental results confirm that the proposed method shows consistent improvement of the performance on low-light datasets.