Most models tasked to ground referential utterances in 2D and 3D scenes learn to select the referred object from a pool of object proposals provided by a pre-trained detector. This is limiting because an utterance may refer to visual entities at various levels of granularity, such as the chair, the leg of the chair, or the tip of the front leg of the chair, which may be missed by the detector. We propose a language grounding model that attends on the referential utterance and on the object proposal pool computed from a pre-trained detector to decode referenced objects with a detection head, without selecting them from the pool. In this way, it is helped by powerful pre-trained object detectors without being restricted by their misses. We call our model Bottom Up Top Down DEtection TRansformers (BUTD-DETR) because it uses both language guidance (top down) and objectness guidance (bottom-up) to ground referential utterances in images and point clouds. Moreover, BUTD-DETR casts object detection as referential grounding and uses object labels as language prompts to be grounded in the visual scene, augmenting supervision for the referential grounding task in this way. The proposed model sets a new state-of-the-art across popular 3D language grounding benchmarks with significant performance gains over previous 3D approaches (12.6% on SR3D, 11.6% on NR3D and 6.3% on ScanRefer). When applied in 2D images, it performs on par with the previous state of the art. We ablate the design choices of our model and quantify their contribution to performance. Our code and checkpoints can be found at the project website https://butd-detr.github.io.
Predicting the object's 6D pose from a single RGB image is a fundamental computer vision task. Generally, the distance between transformed object vertices is employed as an objective function for pose estimation methods. However, projective geometry in the camera space is not considered in those methods and causes performance degradation. In this regard, we propose a new pose estimation system based on a projective grid instead of object vertices. Our pose estimation method, dynamic projective spatial transformer network (DProST), localizes the region of interest grid on the rays in camera space and transforms the grid to object space by estimated pose. The transformed grid is used as both a sampling grid and a new criterion of the estimated pose. Additionally, because DProST does not require object vertices, our method can be used in a mesh-less setting by replacing the mesh with a reconstructed feature. Experimental results show that mesh-less DProST outperforms the state-of-the-art mesh-based methods on the LINEMOD and LINEMOD-OCCLUSION dataset, and shows competitive performance on the YCBV dataset with mesh data. The source code is available at https://github.com/parkjaewoo0611/DProST
Domain adaptive object detection (DAOD) aims to improve the generalization ability of detectors when the training and test data are from different domains. Considering the significant domain gap, some typical methods, e.g., CycleGAN-based methods, adopt the intermediate domain to bridge the source and target domains progressively. However, the CycleGAN-based intermediate domain lacks the pix- or instance-level supervision for object detection, which leads to semantic differences. To address this problem, in this paper, we introduce a Frequency Spectrum Augmentation Consistency (FSAC) framework with four different low-frequency filter operations. In this way, we can obtain a series of augmented data as the intermediate domain. Concretely, we propose a two-stage optimization framework. In the first stage, we utilize all the original and augmented source data to train an object detector. In the second stage, augmented source and target data with pseudo labels are adopted to perform the self-training for prediction consistency. And a teacher model optimized using Mean Teacher is used to further revise the pseudo labels. In the experiment, we evaluate our method on the single- and compound- target DAOD separately, which demonstrate the effectiveness of our method.
Many important tasks are defined in terms of object. To generalize across these tasks, a reinforcement learning (RL) agent needs to exploit the structure that the objects induce. Prior work has either hard-coded object-centric features, used complex object-centric generative models, or updated state using local spatial features. However, these approaches have had limited success in enabling general RL agents. Motivated by this, we introduce "Feature-Attending Recurrent Modules" (FARM), an architecture for learning state representations that relies on simple, broadly applicable inductive biases for capturing spatial and temporal regularities. FARM learns a state representation that is distributed across multiple modules that each attend to spatiotemporal features with an expressive feature attention mechanism. We show that this improves an RL agent's ability to generalize across object-centric tasks. We study task suites in both 2D and 3D environments and find that FARM better generalizes compared to competing architectures that leverage attention or multiple modules.
Existing state-of-the-art methods for Video Object Segmentation (VOS) learn low-level pixel-to-pixel correspondences between frames to propagate object masks across video. This requires a large amount of densely annotated video data, which is costly to annotate, and largely redundant since frames within a video are highly correlated. In light of this, we propose HODOR: a novel method that tackles VOS by effectively leveraging annotated static images for understanding object appearance and scene context. We encode object instances and scene information from an image frame into robust high-level descriptors which can then be used to re-segment those objects in different frames. As a result, HODOR achieves state-of-the-art performance on the DAVIS and YouTube-VOS benchmarks compared to existing methods trained without video annotations. Without any architectural modification, HODOR can also learn from video context around single annotated video frames by utilizing cyclic consistency, whereas other methods rely on dense, temporally consistent annotations. Source code is available at: https://github.com/Ali2500/HODOR
Interactive object understanding, or what we can do to objects and how is a long-standing goal of computer vision. In this paper, we tackle this problem through observation of human hands in in-the-wild egocentric videos. We demonstrate that observation of what human hands interact with and how can provide both the relevant data and the necessary supervision. Attending to hands, readily localizes and stabilizes active objects for learning and reveals places where interactions with objects occur. Analyzing the hands shows what we can do to objects and how. We apply these basic principles on the EPIC-KITCHENS dataset, and successfully learn state-sensitive features, and object affordances (regions of interaction and afforded grasps), purely by observing hands in egocentric videos.
Contrastive language-image pretraining (CLIP) using image-text pairs has achieved impressive results on image classification in both zero-shot and transfer learning settings. However, we show that directly applying such models to recognize image regions for object detection leads to poor performance due to a domain shift: CLIP was trained to match an image as a whole to a text description, without capturing the fine-grained alignment between image regions and text spans. To mitigate this issue, we propose a new method called RegionCLIP that significantly extends CLIP to learn region-level visual representations, thus enabling fine-grained alignment between image regions and textual concepts. Our method leverages a CLIP model to match image regions with template captions and then pretrains our model to align these region-text pairs in the feature space. When transferring our pretrained model to the open-vocabulary object detection tasks, our method significantly outperforms the state of the art by 3.8 AP50 and 2.2 AP for novel categories on COCO and LVIS datasets, respectively. Moreoever, the learned region representations support zero-shot inference for object detection, showing promising results on both COCO and LVIS datasets. Our code is available at https://github.com/microsoft/RegionCLIP.
Human-object interaction (HOI) detection as a downstream of object detection tasks requires localizing pairs of humans and objects and extracting the semantic relationships between humans and objects from an image. Recently, one-stage approaches have become a new trend for this task due to their high efficiency. However, these approaches focus on detecting possible interaction points or filtering human-object pairs, ignoring the variability in the location and size of different objects at spatial scales. To address this problem, we propose a transformer-based method, QAHOI (Query-Based Anchors for Human-Object Interaction detection), which leverages a multi-scale architecture to extract features from different spatial scales and uses query-based anchors to predict all the elements of an HOI instance. We further investigate that a powerful backbone significantly increases accuracy for QAHOI, and QAHOI with a transformer-based backbone outperforms recent state-of-the-art methods by large margins on the HICO-DET benchmark. The source code is available at $\href{https://github.com/cjw2021/QAHOI}{\text{this https URL}}$.
Planar object tracking plays an important role in AI applications, such as robotics, visual servoing, and visual SLAM. Although the previous planar trackers work well in most scenarios, it is still a challenging task due to the rapid motion and large transformation between two consecutive frames. The essential reason behind this problem is that the condition number of such a non-linear system changes unstably when the searching range of the homography parameter space becomes larger. To this end, we propose a novel Homography Decomposition Networks(HDN) approach that drastically reduces and stabilizes the condition number by decomposing the homography transformation into two groups. Specifically, a similarity transformation estimator is designed to predict the first group robustly by a deep convolution equivariant network. By taking advantage of the scale and rotation estimation with high confidence, a residual transformation is estimated by a simple regression model. Furthermore, the proposed end-to-end network is trained in a semi-supervised fashion. Extensive experiments show that our proposed approach outperforms the state-of-the-art planar tracking methods at a large margin on the challenging POT, UCSB and POIC datasets.
3D object detection is a key module for safety-critical robotics applications such as autonomous driving. For these applications, we care most about how the detections affect the ego-agent's behavior and safety (the egocentric perspective). Intuitively, we seek more accurate descriptions of object geometry when it's more likely to interfere with the ego-agent's motion trajectory. However, current detection metrics, based on box Intersection-over-Union (IoU), are object-centric and aren't designed to capture the spatio-temporal relationship between objects and the ego-agent. To address this issue, we propose a new egocentric measure to evaluate 3D object detection, namely Support Distance Error (SDE). Our analysis based on SDE reveals that the egocentric detection quality is bounded by the coarse geometry of the bounding boxes. Given the insight that SDE would benefit from more accurate geometry descriptions, we propose to represent objects as amodal contours, specifically amodal star-shaped polygons, and devise a simple model, StarPoly, to predict such contours. Our experiments on the large-scale Waymo Open Dataset show that SDE better reflects the impact of detection quality on the ego-agent's safety compared to IoU; and the estimated contours from StarPoly consistently improve the egocentric detection quality over recent 3D object detectors.
Recent multi-object tracking (MOT) systems have leveraged highly accurate object detectors; however, training such detectors requires large amounts of labeled data. Although such data is widely available for humans and vehicles, it is significantly more scarce for other animal species. We present Robust Confidence Tracking (RCT), an algorithm designed to maintain robust performance even when detection quality is poor. In contrast to prior methods which discard detection confidence information, RCT takes a fundamentally different approach, relying on the exact detection confidence values to initialize tracks, extend tracks, and filter tracks. In particular, RCT is able to minimize identity switches by efficiently using low-confidence detections (along with a single object tracker) to keep continuous track of objects. To evaluate trackers in the presence of unreliable detections, we present a challenging real-world underwater fish tracking dataset, FISHTRAC. In an evaluation on FISHTRAC as well as the UA-DETRAC dataset, we find that RCT outperforms other algorithms when provided with imperfect detections, including state-of-the-art deep single and multi-object trackers as well as more classic approaches. Specifically, RCT has the best average HOTA across methods that successfully return results for all sequences, and has significantly less identity switches than other methods.
The detection of object states in images (State Detection - SD) is a problem of both theoretical and practical importance and it is tightly interwoven with other important computer vision problems, such as action recognition and affordance detection. It is also highly relevant to any entity that needs to reason and act in dynamic domains, such as robotic systems and intelligent agents. Despite its importance, up to now, the research on this problem has been limited. In this paper, we attempt a systematic study of the SD problem. First, we introduce the Object State Detection Dataset (OSDD), a new publicly available dataset consisting of more than 19,000 annotations for 18 object categories and 9 state classes. Second, using a standard deep learning framework used for Object Detection (OD), we conduct a number of appropriately designed experiments, towards an in-depth study of the behavior of the SD problem. This study enables the setup of a baseline on the performance of SD, as well as its relative performance in comparison to OD, in a variety of scenarios. Overall, the experimental outcomes confirm that SD is harder than OD and that tailored SD methods need to be developed for addressing effectively this significant problem.
Existing studies on salient object detection (SOD) focus on extracting distinct objects with edge information and aggregating multi-level features to improve SOD performance. To achieve satisfactory performance, the methods employ refined edge information and low multi-level discrepancy. However, both performance gain and computational efficiency cannot be attained, which has motivated us to study the inefficiencies in existing encoder-decoder structures to avoid this trade-off. We propose TRACER, which detects salient objects with explicit edges by incorporating attention guided tracing modules. We employ a masked edge attention module at the end of the first encoder using a fast Fourier transform to propagate the refined edge information to the downstream feature extraction. In the multi-level aggregation phase, the union attention module identifies the complementary channel and important spatial information. To improve the decoder performance and computational efficiency, we minimize the decoder block usage with object attention module. This module extracts undetected objects and edge information from refined channels and spatial representations. Subsequently, we propose an adaptive pixel intensity loss function to deal with the relatively important pixels unlike conventional loss functions which treat all pixels equally. A comparison with 13 existing methods reveals that TRACER achieves state-of-the-art performance on five benchmark datasets. We have released TRACER at https://github.com/Karel911/TRACER.
Articulated objects are pervasive in daily life. However, due to the intrinsic high-DoF structure, the joint states of the articulated objects are hard to be estimated. To model articulated objects, two kinds of shape deformations namely the geometric and the pose deformation should be considered. In this work, we present a novel category-specific parametric representation called Object Model with Articulated Deformations (OMAD) to explicitly model the articulated objects. In OMAD, a category is associated with a linear shape function with shared shape basis and a non-linear joint function. Both functions can be learned from a large-scale object model dataset and fixed as category-specific priors. Then we propose an OMADNet to predict the shape parameters and the joint states from an object's single observation. With the full representation of the object shape and joint states, we can address several tasks including category-level object pose estimation and the articulated object retrieval. To evaluate these tasks, we create a synthetic dataset based on PartNet-Mobility. Extensive experiments show that our simple OMADNet can serve as a strong baseline for both tasks.
Deep learning-based approaches have shown remarkable performance in the 3D object detection task. However, they suffer from a catastrophic performance drop on the originally trained classes when incrementally learning new classes without revisiting the old data. This "catastrophic forgetting" phenomenon impedes the deployment of 3D object detection approaches in real-world scenarios, where continuous learning systems are needed. In this paper, we study the unexplored yet important class-incremental 3D object detection problem and present the first solution - SDCoT, a novel static-dynamic co-teaching method. Our SDCoT alleviates the catastrophic forgetting of old classes via a static teacher, which provides pseudo annotations for old classes in the new samples and regularizes the current model by extracting previous knowledge with a distillation loss. At the same time, SDCoT consistently learns the underlying knowledge from new data via a dynamic teacher. We conduct extensive experiments on two benchmark datasets and demonstrate the superior performance of our SDCoT over baseline approaches in several incremental learning scenarios.
In this paper, we propose a new joint object detection and tracking (JoDT) framework for 3D object detection and tracking based on camera and LiDAR sensors. The proposed method, referred to as 3D DetecTrack, enables the detector and tracker to cooperate to generate a spatio-temporal representation of the camera and LiDAR data, with which 3D object detection and tracking are then performed. The detector constructs the spatio-temporal features via the weighted temporal aggregation of the spatial features obtained by the camera and LiDAR fusion. Then, the detector reconfigures the initial detection results using information from the tracklets maintained up to the previous time step. Based on the spatio-temporal features generated by the detector, the tracker associates the detected objects with previously tracked objects using a graph neural network (GNN). We devise a fully-connected GNN facilitated by a combination of rule-based edge pruning and attention-based edge gating, which exploits both spatial and temporal object contexts to improve tracking performance. The experiments conducted on both KITTI and nuScenes benchmarks demonstrate that the proposed 3D DetecTrack achieves significant improvements in both detection and tracking performances over baseline methods and achieves state-of-the-art performance among existing methods through collaboration between the detector and tracker.
Augmented Reality is a promising technique for human-machine interaction. Especially in robotics, which always considers systems in their environment, it is highly beneficial to display visualizations and receive user input directly in exactly that environment. We explore this idea using the Microsoft HoloLens, with which we capture indoor environments and display interaction cues with known object classes. The 3D mesh recorded by the HoloLens is annotated on-line, as the user moves, with semantic classes using a projective approach, which allows us to use a state-of-the-art 2D semantic segmentation method. The results are fused onto the mesh; prominent object segments are identified and displayed in 3D to the user. Finally, the user can trigger actions by gesturing at the object. We both present qualitative results and analyze the accuracy and performance of our method in detail on an indoor dataset.
This paper aims to tackle the challenging task of one-shot object counting. Given an image containing novel, previously unseen category objects, the goal of the task is to count all instances in the desired category with only one supporting bounding box example. To this end, we propose a counting model by which you only need to Look At One instance (LaoNet). First, a feature correlation module combines the Self-Attention and Correlative-Attention modules to learn both inner-relations and inter-relations. It enables the network to be robust to the inconsistency of rotations and sizes among different instances. Second, a Scale Aggregation mechanism is designed to help extract features with different scale information. Compared with existing few-shot counting methods, LaoNet achieves state-of-the-art results while learning with a high convergence speed. The code will be available soon.
Object detection has made tremendous strides in computer vision. Small object detection with appearance degradation is a prominent challenge, especially for aerial observations. To collect sufficient positive/negative samples for heuristic training, most object detectors preset region anchors in order to calculate Intersection-over-Union (IoU) against the ground-truthed data. In this case, small objects are frequently abandoned or mislabeled. In this paper, we present an effective Dynamic Enhancement Anchor (DEA) network to construct a novel training sample generator. Different from the other state-of-the-art techniques, the proposed network leverages a sample discriminator to realize interactive sample screening between an anchor-based unit and an anchor-free unit to generate eligible samples. Besides, multi-task joint training with a conservative anchor-based inference scheme enhances the performance of the proposed model while reducing computational complexity. The proposed scheme supports both oriented and horizontal object detection tasks. Extensive experiments on two challenging aerial benchmarks (i.e., DOTA and HRSC2016) indicate that our method achieves state-of-the-art performance in accuracy with moderate inference speed and computational overhead for training. On DOTA, our DEA-Net which integrated with the baseline of RoI-Transformer surpasses the advanced method by 0.40% mean-Average-Precision (mAP) for oriented object detection with a weaker backbone network (ResNet-101 vs ResNet-152) and 3.08% mean-Average-Precision (mAP) for horizontal object detection with the same backbone. Besides, our DEA-Net which integrated with the baseline of ReDet achieves the state-of-the-art performance by 80.37%. On HRSC2016, it surpasses the previous best model by 1.1% using only 3 horizontal anchors.
We introduce an end-to-end learning framework for image-to-image composition, aiming to plausibly compose an object represented as a cropped patch from an object image into a background scene image. As our approach emphasizes more on semantic and structural coherence of the composed images, rather than their pixel-level RGB accuracies, we tailor the input and output of our network with structure-aware features and design our network losses accordingly, with ground truth established in a self-supervised setting through the object cropping. Specifically, our network takes the semantic layout features from the input scene image, features encoded from the edges and silhouette in the input object patch, as well as a latent code as inputs, and generates a 2D spatial affine transform defining the translation and scaling of the object patch. The learned parameters are further fed into a differentiable spatial transformer network to transform the object patch into the target image, where our model is trained adversarially using an affine transform discriminator and a layout discriminator. We evaluate our network, coined SAC-GAN, for various image composition scenarios in terms of quality, composability, and generalizability of the composite images. Comparisons are made to state-of-the-art alternatives, including Instance Insertion, ST-GAN, CompGAN and PlaceNet, confirming superiority of our method.