In this report, we present our champion solutions to five tracks at Ego4D challenge. We leverage our developed InternVideo, a video foundation model, for five Ego4D tasks, including Moment Queries, Natural Language Queries, Future Hand Prediction, State Change Object Detection, and Short-term Object Interaction Anticipation. InternVideo-Ego4D is an effective paradigm to adapt the strong foundation model to the downstream ego-centric video understanding tasks with simple head designs. In these five tasks, the performance of InternVideo-Ego4D comprehensively surpasses the baseline methods and the champions of CVPR2022, demonstrating the powerful representation ability of InternVideo as a video foundation model. Our code will be released at https://github.com/OpenGVLab/ego4d-eccv2022-solutions
LiDAR and camera, as two different sensors, supply geometric (point clouds) and semantic (RGB images) information of 3D scenes. However, it is still challenging for existing methods to fuse data from the two cross sensors, making them complementary for quality 3D object detection (3OD). We propose ImLiDAR, a new 3OD paradigm to narrow the cross-sensor discrepancies by progressively fusing the multi-scale features of camera Images and LiDAR point clouds. ImLiDAR enables to provide the detection head with cross-sensor yet robustly fused features. To achieve this, two core designs exist in ImLiDAR. First, we propose a cross-sensor dynamic message propagation module to combine the best of the multi-scale image and point features. Second, we raise a direct set prediction problem that allows designing an effective set-based detector to tackle the inconsistency of the classification and localization confidences, and the sensitivity of hand-tuned hyperparameters. Besides, the novel set-based detector can be detachable and easily integrated into various detection networks. Comparisons on both the KITTI and SUN-RGBD datasets show clear visual and numerical improvements of our ImLiDAR over twenty-three state-of-the-art 3OD methods.
Autonomous driving is a popular research area within the computer vision research community. Since autonomous vehicles are highly safety-critical, ensuring robustness is essential for real-world deployment. While several public multimodal datasets are accessible, they mainly comprise two sensor modalities (camera, LiDAR) which are not well suited for adverse weather. In addition, they lack far-range annotations, making it harder to train neural networks that are the base of a highway assistant function of an autonomous vehicle. Therefore, we introduce a multimodal dataset for robust autonomous driving with long-range perception. The dataset consists of 176 scenes with synchronized and calibrated LiDAR, camera, and radar sensors covering a 360-degree field of view. The collected data was captured in highway, urban, and suburban areas during daytime, night, and rain and is annotated with 3D bounding boxes with consistent identifiers across frames. Furthermore, we trained unimodal and multimodal baseline models for 3D object detection. Data are available at \url{https://github.com/aimotive/aimotive_dataset}.
3D object detection from multiple image views is a fundamental and challenging task for visual scene understanding. Owing to its low cost and high efficiency, multi-view 3D object detection has demonstrated promising application prospects. However, accurately detecting objects through perspective views is extremely difficult due to the lack of depth information. Current approaches tend to adopt heavy backbones for image encoders, making them inapplicable for real-world deployment. Different from the images, LiDAR points are superior in providing spatial cues, resulting in highly precise localization. In this paper, we explore the incorporation of LiDAR-based detectors for multi-view 3D object detection. Instead of directly training a depth prediction network, we unify the image and LiDAR features in the Bird-Eye-View (BEV) space and adaptively transfer knowledge across non-homogenous representations in a teacher-student paradigm. To this end, we propose \textbf{BEVDistill}, a cross-modal BEV knowledge distillation (KD) framework for multi-view 3D object detection. Extensive experiments demonstrate that the proposed method outperforms current KD approaches on a highly-competitive baseline, BEVFormer, without introducing any extra cost in the inference phase. Notably, our best model achieves 59.4 NDS on the nuScenes test leaderboard, achieving new state-of-the-art in comparison with various image-based detectors. Code will be available at https://github.com/zehuichen123/BEVDistill.
Nonlinear dimensionality reduction lacks interpretability due to the absence of source features in low-dimensional embedding space. We propose an interpretable method featMAP to preserve source features by tangent space embedding. The core of our proposal is to utilize local singular value decomposition (SVD) to approximate the tangent space which is embedded to low-dimensional space by maintaining the alignment. Based on the embedding tangent space, featMAP enables the interpretability by locally demonstrating the source features and feature importance. Furthermore, featMAP embeds the data points by anisotropic projection to preserve the local similarity and original density. We apply featMAP to interpreting digit classification, object detection and MNIST adversarial examples. FeatMAP uses source features to explicitly distinguish the digits and objects and to explain the misclassification of adversarial examples. We also compare featMAP with other state-of-the-art methods on local and global metrics.
The image-based 3D object detection task expects that the predicted 3D bounding box has a ``tightness'' projection (also referred to as cuboid), which fits the object contour well on the image while still keeping the geometric attribute on the 3D space, e.g., physical dimension, pairwise orthogonal, etc. These requirements bring significant challenges to the annotation. Simply projecting the Lidar-labeled 3D boxes to the image leads to non-trivial misalignment, while directly drawing a cuboid on the image cannot access the original 3D information. In this work, we propose a learning-based 3D box adaptation approach that automatically adjusts minimum parameters of the 360$^{\circ}$ Lidar 3D bounding box to perfectly fit the image appearance of panoramic cameras. With only a few 2D boxes annotation as guidance during the training phase, our network can produce accurate image-level cuboid annotations with 3D properties from Lidar boxes. We call our method ``you only label once'', which means labeling on the point cloud once and automatically adapting to all surrounding cameras. As far as we know, we are the first to focus on image-level cuboid refinement, which balances the accuracy and efficiency well and dramatically reduces the labeling effort for accurate cuboid annotation. Extensive experiments on the public Waymo and NuScenes datasets show that our method can produce human-level cuboid annotation on the image without needing manual adjustment.
The Metaverse has received much attention recently. Metaverse applications via mobile augmented reality (MAR) require rapid and accurate object detection to mix digital data with the real world. Federated learning (FL) is an intriguing distributed machine learning approach due to its privacy-preserving characteristics. Due to privacy concerns and the limited computation resources on mobile devices, we incorporate FL into MAR systems of the Metaverse to train a model cooperatively. Besides, to balance the trade-off between energy, execution latency and model accuracy, thereby accommodating different demands and application scenarios, we formulate an optimization problem to minimize a weighted combination of total energy consumption, completion time and model accuracy. Through decomposing the non-convex optimization problem into two subproblems, we devise a resource allocation algorithm to determine the bandwidth allocation, transmission power, CPU frequency and video frame resolution for each participating device. We further present the convergence analysis and computational complexity of the proposed algorithm. Numerical results show that our proposed algorithm has better performance (in terms of energy consumption, completion time and model accuracy) under different weight parameters compared to existing benchmarks.
Cameras are increasingly being deployed in cities, enterprises and roads world-wide to enable many applications in public safety, intelligent transportation, retail, healthcare and manufacturing. Often, after initial deployment of the cameras, the environmental conditions and the scenes around these cameras change, and our experiments show that these changes can adversely impact the accuracy of insights from video analytics. This is because the camera parameter settings, though optimal at deployment time, are not the best settings for good-quality video capture as the environmental conditions and scenes around a camera change during operation. Capturing poor-quality video adversely affects the accuracy of analytics. To mitigate the loss in accuracy of insights, we propose a novel, reinforcement-learning based system APT that dynamically, and remotely (over 5G networks), tunes the camera parameters, to ensure a high-quality video capture, which mitigates any loss in accuracy of video analytics. As a result, such tuning restores the accuracy of insights when environmental conditions or scene content change. APT uses reinforcement learning, with no-reference perceptual quality estimation as the reward function. We conducted extensive real-world experiments, where we simultaneously deployed two cameras side-by-side overlooking an enterprise parking lot (one camera only has manufacturer-suggested default setting, while the other camera is dynamically tuned by APT during operation). Our experiments demonstrated that due to dynamic tuning by APT, the analytics insights are consistently better at all times of the day: the accuracy of object detection video analytics application was improved on average by ~ 42%. Since our reward function is independent of any analytics task, APT can be readily used for different video analytics tasks.
The quality and size of training sets often limit the performance of many state of the art object detectors. However, in many scenarios, it can be difficult to collect images for training, not to mention the costs associated with collecting annotations suitable for training these object detectors. For these reasons, on challenging video datasets such as the Dataset for Underwater Substrate and Invertebrate Analysis (DUSIA), budgets may only allow for collecting and providing partial annotations. To aid in the challenges associated with training with limited and partial annotations, we introduce Context Matched Collages, which leverage explicit context labels to combine unused background examples with existing annotated data to synthesize additional training samples that ultimately improve object detection performance. By combining a set of our generated collage images with the original training set, we see improved performance using three different object detectors on DUSIA, ultimately achieving state of the art object detection performance on the dataset.
Tables are everywhere, from scientific journals, papers, websites, and newspapers all the way to items we buy at the supermarket. Detecting them is thus of utmost importance to automatically understanding the content of a document. The performance of table detection has substantially increased thanks to the rapid development of deep learning networks. The goals of this survey are to provide a profound comprehension of the major developments in the field of Table Detection, offer insight into the different methodologies, and provide a systematic taxonomy of the different approaches. Furthermore, we provide an analysis of both classic and new applications in the field. Lastly, the datasets and source code of the existing models are organized to provide the reader with a compass on this vast literature. Finally, we go over the architecture of utilizing various object detection and table structure recognition methods to create an effective and efficient system, as well as a set of development trends to keep up with state-of-the-art algorithms and future research. We have also set up a public GitHub repository where we will be updating the most recent publications, open data, and source code. The GitHub repository is available at https://github.com/abdoelsayed2016/table-detection-structure-recognition.
Self-supervised pre-training, based on the pretext task of instance discrimination, has fueled the recent advance in label-efficient object detection. However, existing studies focus on pre-training only a feature extractor network to learn transferable representations for downstream detection tasks. This leads to the necessity of training multiple detection-specific modules from scratch in the fine-tuning phase. We argue that the region proposal network (RPN), a common detection-specific module, can additionally be pre-trained towards reducing the localization error of multi-stage detectors. In this work, we propose a simple pretext task that provides an effective pre-training for the RPN, towards efficiently improving downstream object detection performance. We evaluate the efficacy of our approach on benchmark object detection tasks and additional downstream tasks, including instance segmentation and few-shot detection. In comparison with multi-stage detectors without RPN pre-training, our approach is able to consistently improve downstream task performance, with largest gains found in label-scarce settings.
Wireless tags are increasingly used to track and identify common items of interest such as retail goods, food, medicine, clothing, books, documents, keys, equipment, and more. At the same time, there is a need for labelled visual data featuring such items for the purpose of training object detection and recognition models for robots operating in homes, warehouses, stores, libraries, pharmacies, and so on. In this paper, we ask: can we leverage the tracking and identification capabilities of such tags as a basis for a large-scale automatic image annotation system for robotic perception tasks? We present RF-Annotate, a pipeline for autonomous pixel-wise image annotation which enables robots to collect labelled visual data of objects of interest as they encounter them within their environment. Our pipeline uses unmodified commodity RFID readers and RGB-D cameras, and exploits arbitrary small-scale motions afforded by mobile robotic platforms to spatially map RFIDs to corresponding objects in the scene. Our only assumption is that the objects of interest within the environment are pre-tagged with inexpensive battery-free RFIDs costing 3-15 cents each. We demonstrate the efficacy of our pipeline on several RGB-D sequences of tabletop scenes featuring common objects in a variety of indoor environments.
Despite recent progress in Multiple Object Tracking (MOT), several obstacles such as occlusions, similar objects, and complex scenes remain an open challenge. Meanwhile, a systematic study of the cost-performance tradeoff for the popular tracking-by-detection paradigm is still lacking. This paper introduces SMILEtrack, an innovative object tracker that effectively addresses these challenges by integrating an efficient object detector with a Siamese network-based Similarity Learning Module (SLM). The technical contributions of SMILETrack are twofold. First, we propose an SLM that calculates the appearance similarity between two objects, overcoming the limitations of feature descriptors in Separate Detection and Embedding (SDE) models. The SLM incorporates a Patch Self-Attention (PSA) block inspired by the vision Transformer, which generates reliable features for accurate similarity matching. Second, we develop a Similarity Matching Cascade (SMC) module with a novel GATE function for robust object matching across consecutive video frames, further enhancing MOT performance. Together, these innovations help SMILETrack achieve an improved trade-off between the cost ({\em e.g.}, running speed) and performance (e.g., tracking accuracy) over several existing state-of-the-art benchmarks, including the popular BYTETrack method. SMILETrack outperforms BYTETrack by 0.4-0.8 MOTA and 2.1-2.2 HOTA points on MOT17 and MOT20 datasets. Code is available at https://github.com/pingyang1117/SMILEtrack_Official
It is believed that human vision system (HVS) consists of pre-attentive process and attention process when performing salient object detection (SOD). Based on this fact, we propose a four-stage framework for SOD, in which the first two stages match the \textbf{P}re-\textbf{A}ttentive process consisting of general feature extraction (GFE) and feature preprocessing (FP), and the last two stages are corresponding to \textbf{A}ttention process containing saliency feature extraction (SFE) and the feature aggregation (FA), namely \textbf{PAANet}. According to the pre-attentive process, the GFE stage applies the fully-trained backbone and needs no further finetuning for different datasets. This modification can greatly increase the training speed. The FP stage plays the role of finetuning but works more efficiently because of its simpler structure and fewer parameters. Moreover, in SFE stage we design for saliency feature extraction a novel contrast operator, which works more semantically in contrast with the traditional convolution operator when extracting the interactive information between the foreground and its surroundings. Interestingly, this contrast operator can be cascaded to form a deeper structure and extract higher-order saliency more effective for complex scene. Comparative experiments with the state-of-the-art methods on 5 datasets demonstrate the effectiveness of our framework.
Object detection technology has been widely used in medical field, such as detecting the images of blood cell to count the changes and distribution for assisting the diagnosis of diseases. However, detecting small objects is one of the most challenging and important problems especially in medical scenarios. Most of the objects in medical images are very small but influential. Improving the detection performance of small objects is a very meaningful topic for medical detection. Current researches mainly focus on the extraction of small object features and data augmentation for small object samples, all of these researches focus on extracting the feature space of small objects better. However, in the training process of a detection model, objects of different sizes are mixed together, which may interfere with each other and affect the performance of small object detection. In this paper, we propose a method called pixel level balancing (PLB), which takes into account the number of pixels contained in the detection box as an impact factor to characterize the size of the inspected objects, and uses this as an impact factor. The training loss of each object of different size is adjusted by a weight dynamically, so as to improve the accuracy of small object detection. Finally, through experiments, we demonstrate that the size of objects in object detection interfere with each other. So that we can improve the accuracy of small object detection through PLB operation. This method can perform well with blood cell detection in our experiments.
In recent years, many data augmentation techniques have been proposed to increase the diversity of input data and reduce the risk of overfitting on deep neural networks. In this work, we propose an easy-to-implement and model-free data augmentation method called Local Magnification (LOMA). Different from other geometric data augmentation methods that perform global transformations on images, LOMA generates additional training data by randomly magnifying a local area of the image. This local magnification results in geometric changes that significantly broaden the range of augmentations while maintaining the recognizability of objects. Moreover, we extend the idea of LOMA and random cropping to the feature space to augment the feature map, which further boosts the classification accuracy considerably. Experiments show that our proposed LOMA, though straightforward, can be combined with standard data augmentation to significantly improve the performance on image classification and object detection. And further combination with our feature augmentation techniques, termed LOMA_IF&FO, can continue to strengthen the model and outperform advanced intensity transformation methods for data augmentation.
The great progress of 3D object detectors relies on large-scale data and 3D annotations. The annotation cost for 3D bounding boxes is extremely expensive while the 2D ones are easier and cheaper to collect. In this paper, we introduce a hybrid training framework, enabling us to learn a visual 3D object detector with massive 2D (pseudo) labels, even without 3D annotations. To break through the information bottleneck of 2D clues, we explore a new perspective: Temporal 2D Supervision. We propose a temporal 2D transformation to bridge the 3D predictions with temporal 2D labels. Two steps, including homography wraping and 2D box deduction, are taken to transform the 3D predictions into 2D ones for supervision. Experiments conducted on the nuScenes dataset show strong results (nearly 90% of its fully-supervised performance) with only 25% 3D annotations. We hope our findings can provide new insights for using a large number of 2D annotations for 3D perception.
Recent progress on 2D object detection has featured Cascade RCNN, which capitalizes on a sequence of cascade detectors to progressively improve proposal quality, towards high-quality object detection. However, there has not been evidence in support of building such cascade structures for 3D object detection, a challenging detection scenario with highly sparse LiDAR point clouds. In this work, we present a simple yet effective cascade architecture, named 3D Cascade RCNN, that allocates multiple detectors based on the voxelized point clouds in a cascade paradigm, pursuing higher quality 3D object detector progressively. Furthermore, we quantitatively define the sparsity level of the points within 3D bounding box of each object as the point completeness score, which is exploited as the task weight for each proposal to guide the learning of each stage detector. The spirit behind is to assign higher weights for high-quality proposals with relatively complete point distribution, while down-weight the proposals with extremely sparse points that often incur noise during training. This design of completeness-aware re-weighting elegantly upgrades the cascade paradigm to be better applicable for the sparse input data, without increasing any FLOP budgets. Through extensive experiments on both the KITTI dataset and Waymo Open Dataset, we validate the superiority of our proposed 3D Cascade RCNN, when comparing to state-of-the-art 3D object detection techniques. The source code is publicly available at \url{https://github.com/caiqi/Cascasde-3D}.
3D object detection is a critical task in autonomous driving. Recently multi-modal fusion-based 3D object detection methods, which combine the complementary advantages of LiDAR and camera, have shown great performance improvements over mono-modal methods. However, so far, no methods have attempted to utilize the instance-level contextual image semantics to guide the 3D object detection. In this paper, we propose a simple and effective Painting Adaptive Instance-prior for 3D object detection (PAI3D) to fuse instance-level image semantics flexibly with point cloud features. PAI3D is a multi-modal sequential instance-level fusion framework. It first extracts instance-level semantic information from images, the extracted information, including objects categorical label, point-to-object membership and object position, are then used to augment each LiDAR point in the subsequent 3D detection network to guide and improve detection performance. PAI3D outperforms the state-of-the-art with a large margin on the nuScenes dataset, achieving 71.4 in mAP and 74.2 in NDS on the test split. Our comprehensive experiments show that instance-level image semantics contribute the most to the performance gain, and PAI3D works well with any good-quality instance segmentation models and any modern point cloud 3D encoders, making it a strong candidate for deployment on autonomous vehicles.
Optical coherence tomography (OCT) has become increasingly essential in assisting the treatment of coronary artery disease (CAD). Image-guided solutions such as Percutaneous Coronary Intervention (PCI) are extensively used during the treatment of CAD. However, unidentified calcified regions within a narrowed artery could impair the outcome of the PCI. Prior to treatments, object detection is paramount to automatically procure accurate readings on the location and thickness of calcifications within the artery. Deep learning-based object detection methods have been explored in a variety of applications. The quality of object detection predictions could lead to uncertain results, which are not desirable in safety-critical scenarios. In this work, we implement an object detection model, You-Only-Look-Once v5 (YOLO), on a calcification detection framework within coronary OCT images. We evaluate the uncertainty of predictions based on the expected calibration errors, thus assessing the certainty level of detection results. To calibrate confidence scores of predictions, we implement dependent logistic calibration using each detection result's confidence and center coordinates. With the calibrated confidence score of each prediction, we lower the uncertainty of predictions in calcification detection. Our results show that the YOLO achieves higher precision and recall in comparison with the other object detection model, meanwhile producing more reliable results. The calibrated confidence of prediction results in a confidence error of approximately 0.13, suggesting that the confidence calibration on calcification detection could provide a more trustworthy result, indicating a great potential to assist clinical evaluation of treating the CAD during the imaging-guided procedure.