Contextual information is a valuable cue for Deep Neural Networks (DNNs) to learn better representations and improve accuracy. However, co-occurrence bias in the training dataset may hamper a DNN model's generalizability to unseen scenarios in the real world. For example, in COCO, many object categories have a much higher co-occurrence with men compared to women, which can bias a DNN's prediction in favor of men. Recent works have focused on task-specific training strategies to handle bias in such scenarios, but fixing the available data is often ignored. In this paper, we propose a novel and more generic solution to address the contextual bias in the datasets by selecting a subset of the samples, which is fair in terms of the co-occurrence with various classes for a protected attribute. We introduce a data repair algorithm using the coefficient of variation, which can curate fair and contextually balanced data for a protected class(es). This helps in training a fair model irrespective of the task, architecture or training methodology. Our proposed solution is simple, effective, and can even be used in an active learning setting where the data labels are not present or being generated incrementally. We demonstrate the effectiveness of our algorithm for the task of object detection and multi-label image classification across different datasets. Through a series of experiments, we validate that curating contextually fair data helps make model predictions fair by balancing the true positive rate for the protected class across groups without compromising on the model's overall performance.
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.
Data augmentation reduces the generalization error by forcing a model to learn invariant representations given different transformations of the input image. In computer vision, on top of the standard image processing functions, data augmentation techniques based on regional dropout such as CutOut, MixUp, and CutMix and policy-based selection such as AutoAugment demonstrated state-of-the-art (SOTA) results. With an increasing number of data augmentation algorithms being proposed, the focus is always on optimizing the input-output mapping while not realizing that there might be an untapped value in the transformed images with the same label. We hypothesize that by forcing the representations of two transformations to agree, we can further reduce the model generalization error. We call our proposed method Agreement Maximization or simply AgMax. With this simple constraint applied during training, empirical results show that data augmentation algorithms can further improve the classification accuracy of ResNet50 on ImageNet by up to 1.5%, WideResNet40-2 on CIFAR10 by up to 0.7%, WideResNet40-2 on CIFAR100 by up to 1.6%, and LeNet5 on Speech Commands Dataset by up to 1.4%. Experimental results further show that unlike other regularization terms such as label smoothing, AgMax can take advantage of the data augmentation to consistently improve model generalization by a significant margin. On downstream tasks such as object detection and segmentation on PascalVOC and COCO, AgMax pre-trained models outperforms other data augmentation methods by as much as 1.0mAP (box) and 0.5mAP (mask). Code is available at https://github.com/roatienza/agmax.
Convolutional Neural Networks (CNNs) for visual tasks are believed to learn both the low-level textures and high-level object attributes, throughout the network depth. This paper further investigates the `texture bias' in CNNs. To this end, we regenerate multiple instances of training examples from each original image, through a process we call `repainting'. The repainted examples preserve the shape and structure of the regions and objects within the scenes, but diversify their texture and color. Our method can regenerate a same image at different daylight, season, or weather conditions, can have colorization or de-colorization effects, or even bring back some texture information from blacked-out areas. The in-place repaint allows us to further use these repainted examples for improving the generalization of CNNs. Through an extensive set of experiments, we demonstrate the usefulness of the repainted examples in training, for the tasks of image classification (ImageNet) and object detection (COCO), over several state-of-the-art network architectures at different capacities, and across different data availability regimes.
The diversity of deep learning applications, datasets, and neural network architectures necessitates a careful selection of the architecture and data that match best to a target application. As an attempt to mitigate this dilemma, this paper investigates the idea of combining multiple trained neural networks using unlabeled data. In addition, combining multiple models into one can speed up the inference, result in stronger, more capable models, and allows us to select efficient device-friendly target network architectures. To this end, the proposed method makes use of generation, filtering, and aggregation of reliable pseudo-labels collected from unlabeled data. Our method supports using an arbitrary number of input models with arbitrary architectures and categories. Extensive performance evaluations demonstrated that our method is very effective. For example, for the task of object detection and without using any ground-truth labels, an EfficientDet-D0 trained on Pascal-VOC and an EfficientDet-D1 trained on COCO, can be combined to a RetinaNet-ResNet50 model, with a similar mAP as the supervised training. If fine-tuned in a semi-supervised setting, the combined model achieves +18.6%, +12.6%, and +8.1% mAP improvements over supervised training with 1%, 5%, and 10% of labels.
Wide Area Motion Imagery (WAMI) yields high-resolution images with a large number of extremely small objects. Target objects have large spatial displacements throughout consecutive frames. This nature of WAMI images makes object tracking and detection challenging. In this paper, we present our deep neural network-based combined object detection and tracking model, namely, Heat Map Network (HM-Net). HM-Net is significantly faster than state-of-the-art frame differencing and background subtraction-based methods, without compromising detection and tracking performances. HM-Net follows the object center-based joint detection and tracking paradigm. Simple heat map-based predictions support an unlimited number of simultaneous detections. The proposed method uses two consecutive frames and the object detection heat map obtained from the previous frame as input, which helps HM-Net monitor spatio-temporal changes between frames and keeps track of previously predicted objects. Although reuse of prior object detection heat map acts as a vital feedback-based memory element, it can lead to an unintended surge of false-positive detections. To increase the robustness of the method against false positives and to eliminate low confidence detections, HM-Net employs novel feedback filters and advanced data augmentations. HM-Net outperforms state-of-the-art WAMI moving object detection and tracking methods on the WPAFB dataset with its 96.2% F1 and 94.4% mAP detection scores while achieving a 61.8% mAP tracking score on the same dataset. This performance corresponds to an improvement of 2.1% for F1, 6.1% for mAP scores on detection, and 9.5% for mAP score on tracking over the state-of-the-art.
Knowledge Distillation is becoming one of the primary trends among neural network compression algorithms to improve the generalization performance of a smaller student model with guidance from a larger teacher model. This momentous rise in applications of knowledge distillation is accompanied by the introduction of numerous algorithms for distilling the knowledge such as soft targets and hint layers. Despite this advancement in different techniques for distilling the knowledge, the aggregation of different paths for distillation has not been studied comprehensively. This is of particular significance, not only because different paths have different importance, but also due to the fact that some paths might have negative effects on the generalization performance of the student model. Hence, we need to adaptively adjust the importance of each path to maximize the impact of distillation on the student model. In this paper, we explore different approaches for aggregating these different paths and introduce our proposed adaptive approach based on multitask learning methods. We empirically demonstrate the effectiveness of the proposed approach over other baselines on the applications of knowledge distillation in classification, semantic segmentation, and object detection tasks.
The deployment of Deep Learning (DL) models is still precluded in those contexts where the amount of supervised data is limited. To answer this issue, active learning strategies aim at minimizing the amount of labelled data required to train a DL model. Most active strategies are based on uncertain sample selection, and even often restricted to samples lying close to the decision boundary. These techniques are theoretically sound, but an understanding of the selected samples based on their content is not straightforward, further driving non-experts to consider DL as a black-box. For the first time, here we propose to take into consideration common domain-knowledge and enable non-expert users to train a model with fewer samples. In our Knowledge-driven Active Learning (KAL) framework, rule-based knowledge is converted into logic constraints and their violation is checked as a natural guide for sample selection. We show that even simple relationships among data and output classes offer a way to spot predictions for which the model need supervision. We empirically show that KAL (i) outperforms many active learning strategies, particularly in those contexts where domain knowledge is rich, (ii) it discovers data distribution lying far from the initial training data, (iii) it ensures domain experts that the provided knowledge is acquired by the model, (iv) it is suitable for regression and object recognition tasks unlike uncertainty-based strategies, and (v) its computational demand is low.
Depth is the hallmark of deep neural networks. But more depth means more sequential computation and higher latency. This begs the question -- is it possible to build high-performing "non-deep" neural networks? We show that it is. To do so, we use parallel subnetworks instead of stacking one layer after another. This helps effectively reduce depth while maintaining high performance. By utilizing parallel substructures, we show, for the first time, that a network with a depth of just 12 can achieve top-1 accuracy over 80% on ImageNet, 96% on CIFAR10, and 81% on CIFAR100. We also show that a network with a low-depth (12) backbone can achieve an AP of 48% on MS-COCO. We analyze the scaling rules for our design and show how to increase performance without changing the network's depth. Finally, we provide a proof of concept for how non-deep networks could be used to build low-latency recognition systems. Code is available at https://github.com/imankgoyal/NonDeepNetworks.
3D object detection often involves complicated training and testing pipelines, which require substantial domain knowledge about individual datasets. Inspired by recent non-maximum suppression-free 2D object detection models, we propose a 3D object detection architecture on point clouds. Our method models 3D object detection as message passing on a dynamic graph, generalizing the DGCNN framework to predict a set of objects. In our construction, we remove the necessity of post-processing via object confidence aggregation or non-maximum suppression. To facilitate object detection from sparse point clouds, we also propose a set-to-set distillation approach customized to 3D detection. This approach aligns the outputs of the teacher model and the student model in a permutation-invariant fashion, significantly simplifying knowledge distillation for the 3D detection task. Our method achieves state-of-the-art performance on autonomous driving benchmarks. We also provide abundant analysis of the detection model and distillation framework.
We introduce a framework for multi-camera 3D object detection. In contrast to existing works, which estimate 3D bounding boxes directly from monocular images or use depth prediction networks to generate input for 3D object detection from 2D information, our method manipulates predictions directly in 3D space. Our architecture extracts 2D features from multiple camera images and then uses a sparse set of 3D object queries to index into these 2D features, linking 3D positions to multi-view images using camera transformation matrices. Finally, our model makes a bounding box prediction per object query, using a set-to-set loss to measure the discrepancy between the ground-truth and the prediction. This top-down approach outperforms its bottom-up counterpart in which object bounding box prediction follows per-pixel depth estimation, since it does not suffer from the compounding error introduced by a depth prediction model. Moreover, our method does not require post-processing such as non-maximum suppression, dramatically improving inference speed. We achieve state-of-the-art performance on the nuScenes autonomous driving benchmark.
Studies on self-supervised visual representation learning (SSL) improve encoder backbones to discriminate training samples without labels. While CNN encoders via SSL achieve comparable recognition performance to those via supervised learning, their network attention is under-explored for further improvement. Motivated by the transformers that explore visual attention effectively in recognition scenarios, we propose a CNN Attention REvitalization (CARE) framework to train attentive CNN encoders guided by transformers in SSL. The proposed CARE framework consists of a CNN stream (C-stream) and a transformer stream (T-stream), where each stream contains two branches. C-stream follows an existing SSL framework with two CNN encoders, two projectors, and a predictor. T-stream contains two transformers, two projectors, and a predictor. T-stream connects to CNN encoders and is in parallel to the remaining C-Stream. During training, we perform SSL in both streams simultaneously and use the T-stream output to supervise C-stream. The features from CNN encoders are modulated in T-stream for visual attention enhancement and become suitable for the SSL scenario. We use these modulated features to supervise C-stream for learning attentive CNN encoders. To this end, we revitalize CNN attention by using transformers as guidance. Experiments on several standard visual recognition benchmarks, including image classification, object detection, and semantic segmentation, show that the proposed CARE framework improves CNN encoder backbones to the state-of-the-art performance.
Many emerging applications of intelligent robots need to explore and understand new environments, where it is desirable to detect objects of novel classes on the fly with minimum online efforts. This is an object detection on demand (ODOD) task. It is challenging, because it is impossible to annotate a large number of data on the fly, and the embedded systems are usually unable to perform back-propagation which is essential for training. Most existing few-shot detection methods are confronted here as they need extra training. We propose a novel morphable detector (MD), that simply "morphs" some of its changeable parameters online estimated from the few samples, so as to detect novel classes without any extra training. The MD has two sets of parameters, one for the feature embedding and the other for class representation (called "prototypes"). Each class is associated with a hidden prototype to be learned by integrating the visual and semantic embeddings. The learning of the MD is based on the alternate learning of the feature embedding and the prototypes in an EM-like approach which allows the recovery of an unknown prototype from a few samples of a novel class. Once an MD is learned, it is able to use a few samples of a novel class to directly compute its prototype to fulfill the online morphing process. We have shown the superiority of the MD in Pascal, COCO and FSOD datasets. The code is available https://github.com/Zhaoxiangyun/Morphable-Detector.
Transformers are transforming the landscape of computer vision, especially for recognition tasks. Detection transformers are the first fully end-to-end learning systems for object detection, while vision transformers are the first fully transformer-based architecture for image classification. In this paper, we integrate Vision and Detection Transformers (ViDT) to build an effective and efficient object detector. ViDT introduces a reconfigured attention module to extend the recent Swin Transformer to be a standalone object detector, followed by a computationally efficient transformer decoder that exploits multi-scale features and auxiliary techniques essential to boost the detection performance without much increase in computational load. Extensive evaluation results on the Microsoft COCO benchmark dataset demonstrate that ViDT obtains the best AP and latency trade-off among existing fully transformer-based object detectors, and achieves 49.2AP owing to its high scalability for large models. We will release the code and trained models at https://github.com/naver-ai/vidt
Head detection in the indoor video is an essential component of building occupancy detection. While deep models have achieved remarkable progress in general object detection, they are not satisfying enough in complex indoor scenes. The indoor surveillance video often includes cluttered background objects, among which heads have small scales and diverse poses. In this paper, we propose Motion-aware Pseudo Siamese Network (MPSN), an end-to-end approach that leverages head motion information to guide the deep model to extract effective head features in indoor scenarios. By taking the pixel-wise difference of adjacent frames as the auxiliary input, MPSN effectively enhances human head motion information and removes the irrelevant objects in the background. Compared with prior methods, it achieves superior performance on the two indoor video datasets. Our experiments show that MPSN successfully suppresses static background objects and highlights the moving instances, especially human heads in indoor videos. We also compare different methods to capture head motion, which demonstrates the simplicity and flexibility of MPSN. To validate the robustness of MPSN, we conduct adversarial experiments with a mathematical solution of small perturbations for robust model selection. Finally, for confirming its potential in building control systems, we apply MPSN to occupancy counting. Code is available at https://github.com/pl-share/MPSN.
Humans have the capacity to question what we see and to recognize when our vision is unreliable (e.g., when we realize that we are experiencing a visual illusion). Inspired by this capacity, we present MetaCOG: a hierarchical probabilistic model that can be attached to a neural object detector to monitor its outputs and determine their reliability. MetaCOG achieves this by learning a probabilistic model of the object detector's performance via Bayesian inference -- i.e., a meta-cognitive representation of the network's propensity to hallucinate or miss different object categories. Given a set of video frames processed by an object detector, MetaCOG performs joint inference over the underlying 3D scene and the detector's performance, grounding inference on a basic assumption of object permanence. Paired with three neural object detectors, we show that MetaCOG accurately recovers each detector's performance parameters and improves the overall system's accuracy. We additionally show that MetaCOG is robust to varying levels of error in object detector outputs, showing proof-of-concept for a novel approach to the problem of detecting and correcting errors in vision systems when ground-truth is not available.
Foreign Object Debris (FOD) detection has attracted increased attention in the area of machine learning and computer vision. However, a robust and publicly available image dataset for FOD has not been initialized. To this end, this paper introduces an image dataset of FOD, named FOD in Airports (FOD-A). FOD-A object categories have been selected based on guidance from prior documentation and related research by the Federal Aviation Administration (FAA). In addition to the primary annotations of bounding boxes for object detection, FOD-A provides labeled environmental conditions. As such, each annotation instance is further categorized into three light level categories (bright, dim, and dark) and two weather categories (dry and wet). Currently, FOD-A has released 31 object categories and over 30,000 annotation instances. This paper presents the creation methodology, discusses the publicly available dataset extension process, and demonstrates the practicality of FOD-A with widely used machine learning models for object detection.
Recognizing images with long-tailed distributions remains a challenging problem while there lacks an interpretable mechanism to solve this problem. In this study, we formulate Long-tailed recognition as Domain Adaption (LDA), by modeling the long-tailed distribution as an unbalanced domain and the general distribution as a balanced domain. Within the balanced domain, we propose to slack the generalization error bound, which is defined upon the empirical risks of unbalanced and balanced domains and the divergence between them. We propose to jointly optimize empirical risks of the unbalanced and balanced domains and approximate their domain divergence by intra-class and inter-class distances, with the aim to adapt models trained on the long-tailed distribution to general distributions in an interpretable way. Experiments on benchmark datasets for image recognition, object detection, and instance segmentation validate that our LDA approach, beyond its interpretability, achieves state-of-the-art performance. Code is available at https://github.com/pengzhiliang/LDA.
While building convolutional network-based systems, the toll it takes to train the network is something that cannot be ignored. In cases where we need to append additional capabilities to the existing model, the attention immediately goes towards retraining techniques. In this paper, I show how to leverage knowledge about the dataset to append the class faster while maintaining the speed of inference as well as the accuracies; while reducing the amount of time and data required. The method can extend a class in the existing object detection model in 1/10th of the time compared to the other existing methods. S-Extension patch not only offers faster training but also speed and ease of adaptation, as it can be appended to any existing system, given it fulfills the similarity threshold condition.
Deep learning is a data-hungry approach, which requires massive training data. However, it is time-consuming and labor-intensive to collect abundant fully-annotated training data for all categories. Assuming the existence of base categories with adequate fully-annotated training samples, different paradigms requiring fewer training samples or weaker annotations for novel categories have attracted growing research interest. Among them, zero-shot (resp., few-shot) learning explores using zero (resp., a few) training samples for novel categories, which lowers the quantity requirement for novel categories. Instead, weak-shot learning lowers the quality requirement for novel categories. Specifically, sufficient training samples are collected for novel categories but they only have weak annotations. In different tasks, weak annotations are presented in different forms (e.g., noisy labels for image classification, image labels for object detection, bounding boxes for segmentation), similar to the definitions in weakly supervised learning. Therefore, weak-shot learning can also be treated as weakly supervised learning with auxiliary fully supervised categories. In this paper, we discuss the existing weak-shot learning methodologies in different tasks and summarize the codes at https://github.com/bcmi/Awesome-Weak-Shot-Learning.