Two-stage and query-based instance segmentation methods have achieved remarkable results. However, their segmented masks are still very coarse. In this paper, we present Mask Transfiner for high-quality and efficient instance segmentation. Instead of operating on regular dense tensors, our Mask Transfiner decomposes and represents the image regions as a quadtree. Our transformer-based approach only processes detected error-prone tree nodes and self-corrects their errors in parallel. While these sparse pixels only constitute a small proportion of the total number, they are critical to the final mask quality. This allows Mask Transfiner to predict highly accurate instance masks, at a low computational cost. Extensive experiments demonstrate that Mask Transfiner outperforms current instance segmentation methods on three popular benchmarks, significantly improving both two-stage and query-based frameworks by a large margin of +3.0 mask AP on COCO and BDD100K, and +6.6 boundary AP on Cityscapes. Our code and trained models will be available at http://vis.xyz/pub/transfiner.
The goal of contrastive learning based pre-training is to leverage large quantities of unlabeled data to produce a model that can be readily adapted downstream. Current approaches revolve around solving an image discrimination task: given an anchor image, an augmented counterpart of that image, and some other images, the model must produce representations such that the distance between the anchor and its counterpart is small, and the distances between the anchor and the other images are large. There are two significant problems with this approach: (i) by contrasting representations at the image-level, it is hard to generate detailed object-sensitive features that are beneficial to downstream object-level tasks such as instance segmentation; (ii) the augmentation strategy of producing an augmented counterpart is fixed, making learning less effective at the later stages of pre-training. In this work, we introduce Curricular Contrastive Object-level Pre-training (CCOP) to tackle these problems: (i) we use selective search to find rough object regions and use them to build an inter-image object-level contrastive loss and an intra-image object-level discrimination loss into our pre-training objective; (ii) we present a curriculum learning mechanism that adaptively augments the generated regions, which allows the model to consistently acquire a useful learning signal, even in the later stages of pre-training. Our experiments show that our approach improves on the MoCo v2 baseline by a large margin on multiple object-level tasks when pre-training on multi-object scene image datasets. Code is available at https://github.com/ChenhongyiYang/CCOP.
Open-vocabulary instance segmentation aims at segmenting novel classes without mask annotations. It is an important step toward reducing laborious human supervision. Most existing works first pretrain a model on captioned images covering many novel classes and then finetune it on limited base classes with mask annotations. However, the high-level textual information learned from caption pretraining alone cannot effectively encode the details required for pixel-wise segmentation. To address this, we propose a cross-modal pseudo-labeling framework, which generates training pseudo masks by aligning word semantics in captions with visual features of object masks in images. Thus, our framework is capable of labeling novel classes in captions via their word semantics to self-train a student model. To account for noises in pseudo masks, we design a robust student model that selectively distills mask knowledge by estimating the mask noise levels, hence mitigating the adverse impact of noisy pseudo masks. By extensive experiments, we show the effectiveness of our framework, where we significantly improve mAP score by 4.5% on MS-COCO and 5.1% on the large-scale Open Images & Conceptual Captions datasets compared to the state-of-the-art.
Object-centric representations are a promising path toward more systematic generalization by providing flexible abstractions upon which compositional world models can be built. Recent work on simple 2D and 3D datasets has shown that models with object-centric inductive biases can learn to segment and represent meaningful objects from the statistical structure of the data alone without the need for any supervision. However, such fully-unsupervised methods still fail to scale to diverse realistic data, despite the use of increasingly complex inductive biases such as priors for the size of objects or the 3D geometry of the scene. In this paper, we instead take a weakly-supervised approach and focus on how 1) using the temporal dynamics of video data in the form of optical flow and 2) conditioning the model on simple object location cues can be used to enable segmenting and tracking objects in significantly more realistic synthetic data. We introduce a sequential extension to Slot Attention which we train to predict optical flow for realistic looking synthetic scenes and show that conditioning the initial state of this model on a small set of hints, such as center of mass of objects in the first frame, is sufficient to significantly improve instance segmentation. These benefits generalize beyond the training distribution to novel objects, novel backgrounds, and to longer video sequences. We also find that such initial-state-conditioning can be used during inference as a flexible interface to query the model for specific objects or parts of objects, which could pave the way for a range of weakly-supervised approaches and allow more effective interaction with trained models.
Vehicle classification is a hot computer vision topic, with studies ranging from ground-view up to top-view imagery. In remote sensing, the usage of top-view images allows for understanding city patterns, vehicle concentration, traffic management, and others. However, there are some difficulties when aiming for pixel-wise classification: (a) most vehicle classification studies use object detection methods, and most publicly available datasets are designed for this task, (b) creating instance segmentation datasets is laborious, and (c) traditional instance segmentation methods underperform on this task since the objects are small. Thus, the present research objectives are: (1) propose a novel semi-supervised iterative learning approach using GIS software, (2) propose a box-free instance segmentation approach, and (3) provide a city-scale vehicle dataset. The iterative learning procedure considered: (1) label a small number of vehicles, (2) train on those samples, (3) use the model to classify the entire image, (4) convert the image prediction into a polygon shapefile, (5) correct some areas with errors and include them in the training data, and (6) repeat until results are satisfactory. To separate instances, we considered vehicle interior and vehicle borders, and the DL model was the U-net with the Efficient-net-B7 backbone. When removing the borders, the vehicle interior becomes isolated, allowing for unique object identification. To recover the deleted 1-pixel borders, we proposed a simple method to expand each prediction. The results show better pixel-wise metrics when compared to the Mask-RCNN (82% against 67% in IoU). On per-object analysis, the overall accuracy, precision, and recall were greater than 90%. This pipeline applies to any remote sensing target, being very efficient for segmentation and generating datasets.
Estimating solar rooftop potential at a national level is a fundamental building block for every country to utilize solar power efficiently. Solar rooftop potential assessment relies on several features such as building geometry, location, and surrounding facilities. Hence, national-level approximations that do not take these factors into deep consideration are often inaccurate. This paper introduces Lebanon's first comprehensive footprint and solar rooftop potential maps using deep learning-based instance segmentation to extract buildings' footprints from satellite images. A photovoltaic panels placement algorithm that considers the morphology of each roof is proposed. We show that the average rooftop's solar potential can fulfill the yearly electric needs of a single-family residence while using only 5% of the roof surface. The usage of 50% of a residential apartment rooftop area would achieve energy security for up to 8 households. We also compute the average and total solar rooftop potential per district to localize regions corresponding to the highest and lowest solar rooftop potential yield. Factors such as size, ground coverage ratio and PV_out are carefully investigated for each district. Baalbeck district yielded the highest total solar rooftop potential despite its low built-up area. While, Beirut capital city has the highest average solar rooftop potential due to its extremely populated urban nature. Reported results and analysis reveal solar rooftop potential urban patterns and provides policymakers and key stakeholders with tangible insights. Lebanon's total solar rooftop potential is about 28.1 TWh/year, two times larger than the national energy consumption in 2019.
In this paper, we present a novel deep neural network architecture for joint class-agnostic object segmentation and grasp detection for robotic picking tasks using a parallel-plate gripper. We introduce depth-aware Coordinate Convolution (CoordConv), a method to increase accuracy for point proposal based object instance segmentation in complex scenes without adding any additional network parameters or computation complexity. Depth-aware CoordConv uses depth data to extract prior information about the location of an object to achieve highly accurate object instance segmentation. These resulting segmentation masks, combined with predicted grasp candidates, lead to a complete scene description for grasping using a parallel-plate gripper. We evaluate the accuracy of grasp detection and instance segmentation on challenging robotic picking datasets, namely Sil\'eane and OCID_grasp, and show the benefit of joint grasp detection and segmentation on a real-world robotic picking task.
The task of instance segmentation in remote sensing images, aiming at performing per-pixel labeling of objects at instance level, is of great importance for various civil applications. Despite previous successes, most existing instance segmentation methods designed for natural images encounter sharp performance degradations when they are directly applied to top-view remote sensing images. Through careful analysis, we observe that the challenges mainly come from the lack of discriminative object features due to severe scale variations, low contrasts, and clustered distributions. In order to address these problems, a novel context aggregation network (CATNet) is proposed to improve the feature extraction process. The proposed model exploits three lightweight plug-and-play modules, namely dense feature pyramid network (DenseFPN), spatial context pyramid (SCP), and hierarchical region of interest extractor (HRoIE), to aggregate global visual context at feature, spatial, and instance domains, respectively. DenseFPN is a multi-scale feature propagation module that establishes more flexible information flows by adopting inter-level residual connections, cross-level dense connections, and feature re-weighting strategy. Leveraging the attention mechanism, SCP further augments the features by aggregating global spatial context into local regions. For each instance, HRoIE adaptively generates RoI features for different downstream tasks. Extensive evaluations of the proposed scheme on iSAID, DIOR, NWPU VHR-10, and HRSID datasets demonstrate that the proposed approach outperforms state-of-the-arts under similar computational costs. Source code and pre-trained models are available at https://github.com/yeliudev/CATNet.
Recently, self-supervised methods show remarkable achievements in image-level representation learning. Nevertheless, their image-level self-supervisions lead the learned representation to sub-optimal for dense prediction tasks, such as object detection, instance segmentation, etc. To tackle this issue, several recent self-supervised learning methods have extended image-level single embedding to pixel-level dense embeddings. Unlike image-level representation learning, due to the spatial deformation of augmentation, it is difficult to sample pixel-level positive pairs. Previous studies have sampled pixel-level positive pairs using the winner-takes-all among similarity or thresholding warped distance between dense embeddings. However, these naive methods can be struggled by background clutter and outliers problems. In this paper, we introduce Hough Contrastive Learning (HoughCL), a Hough space based method that enforces geometric consistency between two dense features. HoughCL achieves robustness against background clutter and outliers. Furthermore, compared to baseline, our dense positive pairing method has no additional learnable parameters and has a small extra computation cost. Compared to previous works, our method shows better or comparable performance on dense prediction fine-tuning tasks.
Paris-CARLA-3D is a dataset of several dense colored point clouds of outdoor environments built by a mobile LiDAR and camera system. The data are composed of two sets with synthetic data from the open source CARLA simulator (700 million points) and real data acquired in the city of Paris (60 million points), hence the name Paris-CARLA-3D. One of the advantages of this dataset is to have simulated the same LiDAR and camera platform in the open source CARLA simulator as the one used to produce the real data. In addition, manual annotation of the classes using the semantic tags of CARLA was performed on the real data, allowing the testing of transfer methods from the synthetic to the real data. The objective of this dataset is to provide a challenging dataset to evaluate and improve methods on difficult vision tasks for the 3D mapping of outdoor environments: semantic segmentation, instance segmentation, and scene completion. For each task, we describe the evaluation protocol as well as the experiments carried out to establish a baseline.
Image segmentation for video analysis plays an essential role in different research fields such as smart city, healthcare, computer vision and geoscience, and remote sensing applications. In this regard, a significant effort has been devoted recently to developing novel segmentation strategies; one of the latest outstanding achievements is panoptic segmentation. The latter has resulted from the fusion of semantic and instance segmentation. Explicitly, panoptic segmentation is currently under study to help gain a more nuanced knowledge of the image scenes for video surveillance, crowd counting, self-autonomous driving, medical image analysis, and a deeper understanding of the scenes in general. To that end, we present in this paper the first comprehensive review of existing panoptic segmentation methods to the best of the authors' knowledge. Accordingly, a well-defined taxonomy of existing panoptic techniques is performed based on the nature of the adopted algorithms, application scenarios, and primary objectives. Moreover, the use of panoptic segmentation for annotating new datasets by pseudo-labeling is discussed. Moving on, ablation studies are carried out to understand the panoptic methods from different perspectives. Moreover, evaluation metrics suitable for panoptic segmentation are discussed, and a comparison of the performance of existing solutions is provided to inform the state-of-the-art and identify their limitations and strengths. Lastly, the current challenges the subject technology faces and the future trends attracting considerable interest in the near future are elaborated, which can be a starting point for the upcoming research studies. The papers provided with code are available at: https://github.com/elharroussomar/Awesome-Panoptic-Segmentation
Large-scale NLP models have been shown to significantly improve the performance on language tasks with no signs of saturation. They also demonstrate amazing few-shot capabilities like that of human beings. This paper aims to explore large-scale models in computer vision. We tackle three major issues in training and application of large vision models, including training instability, resolution gaps between pre-training and fine-tuning, and hunger on labelled data. Three main techniques are proposed: 1) a residual-post-norm method combined with cosine attention to improve training stability; 2) A log-spaced continuous position bias method to effectively transfer models pre-trained using low-resolution images to downstream tasks with high-resolution inputs; 3) A self-supervised pre-training method, SimMIM, to reduce the needs of vast labeled images. Through these techniques, this paper successfully trained a 3 billion-parameter Swin Transformer V2 model, which is the largest dense vision model to date, and makes it capable of training with images of up to 1,536$\times$1,536 resolution. It set new performance records on 4 representative vision tasks, including ImageNet-V2 image classification, COCO object detection, ADE20K semantic segmentation, and Kinetics-400 video action classification. Also note our training is much more efficient than that in Google's billion-level visual models, which consumes 40 times less labelled data and 40 times less training time. Code is available at \url{https://github.com/microsoft/Swin-Transformer}.
Mixup-based augmentation has been found to be effective for generalizing models during training, especially for Vision Transformers (ViTs) since they can easily overfit. However, previous mixup-based methods have an underlying prior knowledge that the linearly interpolated ratio of targets should be kept the same as the ratio proposed in input interpolation. This may lead to a strange phenomenon that sometimes there is no valid object in the mixed image due to the random process in augmentation but there is still response in the label space. To bridge such gap between the input and label spaces, we propose TransMix, which mixes labels based on the attention maps of Vision Transformers. The confidence of the label will be larger if the corresponding input image is weighted higher by the attention map. TransMix is embarrassingly simple and can be implemented in just a few lines of code without introducing any extra parameters and FLOPs to ViT-based models. Experimental results show that our method can consistently improve various ViT-based models at scales on ImageNet classification. After pre-trained with TransMix on ImageNet, the ViT-based models also demonstrate better transferability to semantic segmentation, object detection and instance segmentation. TransMix also exhibits to be more robust when evaluating on 4 different benchmarks. Code will be made publicly available at https://github.com/Beckschen/TransMix.
3D semantic segmentation is a fundamental building block for several scene understanding applications such as autonomous driving, robotics and AR/VR. Several state-of-the-art semantic segmentation models suffer from the part misclassification problem, wherein parts of the same object are labelled incorrectly. Previous methods have utilized hierarchical, iterative methods to fuse semantic and instance information, but they lack learnability in context fusion, and are computationally complex and heuristic driven. This paper presents Segment-Fusion, a novel attention-based method for hierarchical fusion of semantic and instance information to address the part misclassifications. The presented method includes a graph segmentation algorithm for grouping points into segments that pools point-wise features into segment-wise features, a learnable attention-based network to fuse these segments based on their semantic and instance features, and followed by a simple yet effective connected component labelling algorithm to convert segment features to instance labels. Segment-Fusion can be flexibly employed with any network architecture for semantic/instance segmentation. It improves the qualitative and quantitative performance of several semantic segmentation backbones by upto 5% when evaluated on the ScanNet and S3DIS datasets.
Although deep learning methods have achieved advanced video object recognition performance in recent years, perceiving heavily occluded objects in a video is still a very challenging task. To promote the development of occlusion understanding, we collect a large-scale dataset called OVIS for video instance segmentation in the occluded scenario. OVIS consists of 296k high-quality instance masks and 901 occluded scenes. While our human vision systems can perceive those occluded objects by contextual reasoning and association, our experiments suggest that current video understanding systems cannot. On the OVIS dataset, all baseline methods encounter a significant performance degradation of about 80% in the heavily occluded object group, which demonstrates that there is still a long way to go in understanding obscured objects and videos in a complex real-world scenario. To facilitate the research on new paradigms for video understanding systems, we launched a challenge based on the OVIS dataset. The submitted top-performing algorithms have achieved much higher performance than our baselines. In this paper, we will introduce the OVIS dataset and further dissect it by analyzing the results of baselines and submitted methods. The OVIS dataset and challenge information can be found at http://songbai.site/ovis .
The success of language Transformers is primarily attributed to the pretext task of masked language modeling (MLM), where texts are first tokenized into semantically meaningful pieces. In this work, we study masked image modeling (MIM) and indicate the advantages and challenges of using a semantically meaningful visual tokenizer. We present a self-supervised framework iBOT that can perform masked prediction with an online tokenizer. Specifically, we perform self-distillation on masked patch tokens and take the teacher network as the online tokenizer, along with self-distillation on the class token to acquire visual semantics. The online tokenizer is jointly learnable with the MIM objective and dispenses with a multi-stage training pipeline where the tokenizer needs to be pre-trained beforehand. We show the prominence of iBOT by achieving an 82.3% linear probing accuracy and an 87.8% fine-tuning accuracy evaluated on ImageNet-1K. Beyond the state-of-the-art image classification results, we underline emerging local semantic patterns, which helps the models to obtain strong robustness against common corruptions and achieve leading results on dense downstream tasks, eg., object detection, instance segmentation, and semantic segmentation.
Video instance segmentation aims to detect, segment, and track objects in a video. Current approaches extend image-level segmentation algorithms to the temporal domain. However, this results in temporally inconsistent masks. In this work, we identify the mask quality due to temporal stability as a performance bottleneck. Motivated by this, we propose a video instance segmentation method that alleviates the problem due to missing detections. Since this cannot be solved simply using spatial information, we leverage temporal context using inter-frame attentions. This allows our network to refocus on missing objects using box predictions from the neighbouring frame, thereby overcoming missing detections. Our method significantly outperforms previous state-of-the-art algorithms using the Mask R-CNN backbone, by achieving 36.0% mAP on the YouTube-VIS benchmark. Additionally, our method is completely online and requires no future frames. Our code is publicly available at https://github.com/anirudh-chakravarthy/ObjProp.
We present deepflash2, a deep learning solution that facilitates the objective and reliable segmentation of ambiguous bioimages through multi-expert annotations and integrated quality assurance. Thereby, deepflash2 addresses typical challenges that arise during training, evaluation, and application of deep learning models in bioimaging. The tool is embedded in an easy-to-use graphical user interface and offers best-in-class predictive performance for semantic and instance segmentation under economical usage of computational resources.
Cell individualization has a vital role in digital pathology image analysis. Deep Learning is considered as an efficient tool for instance segmentation tasks, including cell individualization. However, the precision of the Deep Learning model relies on massive unbiased dataset and manual pixel-level annotations, which is labor intensive. Moreover, most applications of Deep Learning have been developed for processing oncological data. To overcome these challenges, i) we established a pipeline to synthesize pixel-level labels with only point annotations provided; ii) we tested an ensemble Deep Learning algorithm to perform cell individualization on neurological data. Results suggest that the proposed method successfully segments neuronal cells in both object-level and pixel-level, with an average detection accuracy of 0.93.
Spiking Neural Networks (SNN) and the field of Neuromorphic Engineering has brought about a paradigm shift in how to approach Machine Learning (ML) and Computer Vision (CV) problem. This paradigm shift comes from the adaption of event-based sensing and processing. An event-based vision sensor allows for sparse and asynchronous events to be produced that are dynamically related to the scene. Allowing not only the spatial information but a high-fidelity of temporal information to be captured. Meanwhile avoiding the extra overhead and redundancy of conventional high frame rate approaches. However, with this change in paradigm, many techniques from traditional CV and ML are not applicable to these event-based spatial-temporal visual streams. As such a limited number of recognition, detection and segmentation approaches exist. In this paper, we present a novel approach that can perform instance segmentation using just the weights of a Spike Time Dependent Plasticity trained Spiking Convolutional Neural Network that was trained for object recognition. This exploits the spatial and temporal aspects of the network's internal feature representations adding this new discriminative capability. We highlight the new capability by successfully transforming a single class unsupervised network for face detection into a multi-person face recognition and instance segmentation network.