Although increasingly training-expensive, most self-supervised learning (SSL) models have repeatedly been trained from scratch but not fully utilized, since only a few SOTAs are employed for downstream tasks. In this work, we explore a sustainable SSL framework with two major challenges: i) learning a stronger new SSL model based on the existing pretrained SSL model, also called as "base" model, in a cost-friendly manner, ii) allowing the training of the new model to be compatible with various base models. We propose a Target-Enhanced Conditional (TEC) scheme which introduces two components to the existing mask-reconstruction based SSL. Firstly, we propose patch-relation enhanced targets which enhances the target given by base model and encourages the new model to learn semantic-relation knowledge from the base model by using incomplete inputs. This hardening and target-enhancing help the new model surpass the base model, since they enforce additional patch relation modeling to handle incomplete input. Secondly, we introduce a conditional adapter that adaptively adjusts new model prediction to align with the target of different base models. Extensive experimental results show that our TEC scheme can accelerate the learning speed, and also improve SOTA SSL base models, e.g., MAE and iBOT, taking an explorative step towards sustainable SSL.
A mainstream type of current self-supervised learning methods pursues a general-purpose representation that can be well transferred to downstream tasks, typically by optimizing on a given pretext task such as instance discrimination. In this work, we argue that existing pretext tasks inevitably introduce biases into the learned representation, which in turn leads to biased transfer performance on various downstream tasks. To cope with this issue, we propose Maximum Entropy Coding (MEC), a more principled objective that explicitly optimizes on the structure of the representation, so that the learned representation is less biased and thus generalizes better to unseen downstream tasks. Inspired by the principle of maximum entropy in information theory, we hypothesize that a generalizable representation should be the one that admits the maximum entropy among all plausible representations. To make the objective end-to-end trainable, we propose to leverage the minimal coding length in lossy data coding as a computationally tractable surrogate for the entropy, and further derive a scalable reformulation of the objective that allows fast computation. Extensive experiments demonstrate that MEC learns a more generalizable representation than previous methods based on specific pretext tasks. It achieves state-of-the-art performance consistently on various downstream tasks, including not only ImageNet linear probe, but also semi-supervised classification, object detection, instance segmentation, and object tracking. Interestingly, we show that existing batch-wise and feature-wise self-supervised objectives could be seen equivalent to low-order approximations of MEC. Code and pre-trained models are available at https://github.com/xinliu20/MEC.
The recent progress in self-supervised learning has successfully combined Masked Image Modeling (MIM) with Siamese Networks, harnessing the strengths of both methodologies. Nonetheless, certain challenges persist when integrating conventional erase-based masking within Siamese ConvNets. Two primary concerns are: (1) The continuous data processing nature of ConvNets, which doesn't allow for the exclusion of non-informative masked regions, leading to reduced training efficiency compared to ViT architecture; (2) The misalignment between erase-based masking and the contrastive-based objective, distinguishing it from the MIM technique. To address these challenges, this work introduces a novel filling-based masking approach, termed \textbf{MixMask}. The proposed method replaces erased areas with content from a different image, effectively countering the information depletion seen in traditional masking methods. Additionally, we unveil an adaptive loss function that captures the semantics of the newly patched views, ensuring seamless integration within the architectural framework. We empirically validate the effectiveness of our approach through comprehensive experiments across various datasets and application scenarios. The findings underscore our framework's enhanced performance in areas such as linear probing, semi-supervised and supervised finetuning, object detection and segmentation. Notably, our method surpasses the MSCN, establishing MixMask as a more advantageous masking solution for Siamese ConvNets. Our code and models are publicly available at https://github.com/kirill-vish/MixMask.
We tackle the problem of novel class discovery and localization (NCDL). In this setting, we assume a source dataset with supervision for only some object classes. Instances of other classes need to be discovered, classified, and localized automatically based on visual similarity without any human supervision. To tackle NCDL, we propose a two-stage object detection network Region-based NCDL (RNCDL) that uses a region proposal network to localize regions of interest (RoIs). We then train our network to learn to classify each RoI, either as one of the known classes, seen in the source dataset, or one of the novel classes, with a long-tail distribution constraint on the class assignments, reflecting the natural frequency of classes in the real world. By training our detection network with this objective in an end-to-end manner, it learns to classify all region proposals for a large variety of classes, including those not part of the labeled object class vocabulary. Our experiments conducted using COCO and LVIS datasets reveal that our method is significantly more effective than multi-stage pipelines that rely on traditional clustering algorithms. Furthermore, we demonstrate the generality of our approach by applying our method to a large-scale Visual Genome dataset, where our network successfully learns to detect various semantic classes without direct supervision.
Masked Image Modeling (MIM) has recently been established as a potent pre-training paradigm. A pretext task is constructed by masking patches in an input image, and this masked content is then predicted by a neural network using visible patches as sole input. This pre-training leads to state-of-the-art performance when finetuned for high-level semantic tasks, e.g. image classification and object detection. In this paper we instead seek to learn representations that transfer well to a wide variety of 3D vision and lower-level geometric downstream tasks, such as depth prediction or optical flow estimation. Inspired by MIM, we propose an unsupervised representation learning task trained from pairs of images showing the same scene from different viewpoints. More precisely, we propose the pretext task of cross-view completion where the first input image is partially masked, and this masked content has to be reconstructed from the visible content and the second image. In single-view MIM, the masked content often cannot be inferred precisely from the visible portion only, so the model learns to act as a prior influenced by high-level semantics. In contrast, this ambiguity can be resolved with cross-view completion from the second unmasked image, on the condition that the model is able to understand the spatial relationship between the two images. Our experiments show that our pretext task leads to significantly improved performance for monocular 3D vision downstream tasks such as depth estimation. In addition, our model can be directly applied to binocular downstream tasks like optical flow or relative camera pose estimation, for which we obtain competitive results without bells and whistles, i.e., using a generic architecture without any task-specific design.
Generating 3D point cloud (PC) data from noisy sonar measurements is a problem that has potential applications for bathymetry mapping, artificial object inspection, mapping of aquatic plants and fauna as well as underwater navigation and localization of vehicles such as submarines. Side-scan sonar sensors are available in inexpensive cost ranges, especially in fish-finders, where the transducers are usually mounted to the bottom of a boat and can approach shallower depths than the ones attached to an Uncrewed Underwater Vehicle (UUV) can. However, extracting 3D information from side-scan sonar imagery is a difficult task because of its low signal-to-noise ratio and missing angle and depth information in the imagery. Since most algorithms that generate a 3D point cloud from side-scan sonar imagery use Shape from Shading (SFS) techniques, extracting 3D information is especially difficult when the seafloor is smooth, is slowly changing in depth, or does not have identifiable objects that make acoustic shadows. This paper introduces an efficient algorithm that generates a sparse 3D point cloud from side-scan sonar images. This computation is done in a computationally efficient manner by leveraging the geometry of the first sonar return combined with known positions provided by GPS and down-scan sonar depth measurement at each data point. Additionally, this paper implements another algorithm that uses a Convolutional Neural Network (CNN) using transfer learning to perform object detection on side-scan sonar images collected in real life and generated with a simulation. The algorithm was tested on both real and synthetic images to show reasonably accurate anomaly detection and classification.
Wrong-way driving is one of the main causes of road accidents and traffic jam all over the world. By detecting wrong-way vehicles, the number of accidents can be minimized and traffic jam can be reduced. With the increasing popularity of real-time traffic management systems and due to the availability of cheaper cameras, the surveillance video has become a big source of data. In this paper, we propose an automatic wrong-way vehicle detection system from on-road surveillance camera footage. Our system works in three stages: the detection of vehicles from the video frame by using the You Only Look Once (YOLO) algorithm, track each vehicle in a specified region of interest using centroid tracking algorithm and detect the wrong-way driving vehicles. YOLO is very accurate in object detection and the centroid tracking algorithm can track any moving object efficiently. Experiment with some traffic videos shows that our proposed system can detect and identify any wrong-way vehicle in different light and weather conditions. The system is very simple and easy to implement.
Detecting occluded objects still remains a challenge for state-of-the-art object detectors. The objective of this work is to improve the detection for such objects, and thereby improve the overall performance of a modern object detector. To this end we make the following four contributions: (1) We propose a simple 'plugin' module for the detection head of two-stage object detectors to improve the recall of partially occluded objects. The module predicts a tri-layer of segmentation masks for the target object, the occluder and the occludee, and by doing so is able to better predict the mask of the target object. (2) We propose a scalable pipeline for generating training data for the module by using amodal completion of existing object detection and instance segmentation training datasets to establish occlusion relationships. (3) We also establish a COCO evaluation dataset to measure the recall performance of partially occluded and separated objects. (4) We show that the plugin module inserted into a two-stage detector can boost the performance significantly, by only fine-tuning the detection head, and with additional improvements if the entire architecture is fine-tuned. COCO results are reported for Mask R-CNN with Swin-T or Swin-S backbones, and Cascade Mask R-CNN with a Swin-B backbone.
Existing multi-agent perception algorithms usually select to share deep neural features extracted from raw sensing data between agents, achieving a trade-off between accuracy and communication bandwidth limit. However, these methods assume all agents have identical neural networks, which might not be practical in the real world. The transmitted features can have a large domain gap when the models differ, leading to a dramatic performance drop in multi-agent perception. In this paper, we propose the first lightweight framework to bridge such domain gaps for multi-agent perception, which can be a plug-in module for most existing systems while maintaining confidentiality. Our framework consists of a learnable feature resizer to align features in multiple dimensions and a sparse cross-domain transformer for domain adaption. Extensive experiments on the public multi-agent perception dataset V2XSet have demonstrated that our method can effectively bridge the gap for features from different domains and outperform other baseline methods significantly by at least 8% for point-cloud-based 3D object detection.
This paper surveys vision-language pre-training (VLP) methods for multimodal intelligence that have been developed in the last few years. We group these approaches into three categories: ($i$) VLP for image-text tasks, such as image captioning, image-text retrieval, visual question answering, and visual grounding; ($ii$) VLP for core computer vision tasks, such as (open-set) image classification, object detection, and segmentation; and ($iii$) VLP for video-text tasks, such as video captioning, video-text retrieval, and video question answering. For each category, we present a comprehensive review of state-of-the-art methods, and discuss the progress that has been made and challenges still being faced, using specific systems and models as case studies. In addition, for each category, we discuss advanced topics being actively explored in the research community, such as big foundation models, unified modeling, in-context few-shot learning, knowledge, robustness, and computer vision in the wild, to name a few.
Self-supervised learning (SSL) methods aim to exploit the abundance of unlabelled data for machine learning (ML), however the underlying principles are often method-specific. An SSL framework derived from biological first principles of embodied learning could unify the various SSL methods, help elucidate learning in the brain, and possibly improve ML. SSL commonly transforms each training datapoint into a pair of views, uses the knowledge of this pairing as a positive (i.e. non-contrastive) self-supervisory sign, and potentially opposes it to unrelated, (i.e. contrastive) negative examples. Here, we show that this type of self-supervision is an incomplete implementation of a concept from neuroscience, the Efference Copy (EC). Specifically, the brain also transforms the environment through efference, i.e. motor commands, however it sends to itself an EC of the full commands, i.e. more than a mere SSL sign. In addition, its action representations are likely egocentric. From such a principled foundation we formally recover and extend SSL methods such as SimCLR, BYOL, and ReLIC under a common theoretical framework, i.e. Self-supervision Through Efference Copies (S-TEC). Empirically, S-TEC restructures meaningfully the within- and between-class representations. This manifests as improvement in recent strong SSL baselines in image classification, segmentation, object detection, and in audio. These results hypothesize a testable positive influence from the brain's motor outputs onto its sensory representations.
Knowledge distillation learns a lightweight student model that mimics a cumbersome teacher. Existing methods regard the knowledge as the feature of each instance or their relations, which is the instance-level knowledge only from the teacher model, i.e., the local knowledge. However, the empirical studies show that the local knowledge is much noisy in object detection tasks, especially on the blurred, occluded, or small instances. Thus, a more intrinsic approach is to measure the representations of instances w.r.t. a group of common basis vectors in the two feature spaces of the teacher and the student detectors, i.e., global knowledge. Then, the distilling algorithm can be applied as space alignment. To this end, a novel prototype generation module (PGM) is proposed to find the common basis vectors, dubbed prototypes, in the two feature spaces. Then, a robust distilling module (RDM) is applied to construct the global knowledge based on the prototypes and filtrate noisy global and local knowledge by measuring the discrepancy of the representations in two feature spaces. Experiments with Faster-RCNN and RetinaNet on PASCAL and COCO datasets show that our method achieves the best performance for distilling object detectors with various backbones, which even surpasses the performance of the teacher model. We also show that the existing methods can be easily combined with global knowledge and obtain further improvement. Code is available: https://github.com/hikvision-research/DAVAR-Lab-ML.
We introduce MOVE, a novel method to segment objects without any form of supervision. MOVE exploits the fact that foreground objects can be shifted locally relative to their initial position and result in realistic (undistorted) new images. This property allows us to train a segmentation model on a dataset of images without annotation and to achieve state of the art (SotA) performance on several evaluation datasets for unsupervised salient object detection and segmentation. In unsupervised single object discovery, MOVE gives an average CorLoc improvement of 7.2% over the SotA, and in unsupervised class-agnostic object detection it gives a relative AP improvement of 53% on average. Our approach is built on top of self-supervised features (e.g. from DINO or MAE), an inpainting network (based on the Masked AutoEncoder) and adversarial training.
LiDAR 3D object detection models are inevitably biased towards their training dataset. The detector clearly exhibits this bias when employed on a target dataset, particularly towards object sizes. However, object sizes vary heavily between domains due to, for instance, different labeling policies or geographical locations. State-of-the-art unsupervised domain adaptation approaches outsource methods to overcome the object size bias. Mainstream size adaptation approaches exploit target domain statistics, contradicting the original unsupervised assumption. Our novel unsupervised anchor calibration method addresses this limitation. Given a model trained on the source data, we estimate the optimal target anchors in a completely unsupervised manner. The main idea stems from an intuitive observation: by varying the anchor sizes for the target domain, we inevitably introduce noise or even remove valuable object cues. The latent object representation, perturbed by the anchor size, is closest to the learned source features only under the optimal target anchors. We leverage this observation for anchor size optimization. Our experimental results show that, without any retraining, we achieve competitive results even compared to state-of-the-art weakly-supervised size adaptation approaches. In addition, our anchor calibration can be combined with such existing methods, making them completely unsupervised.
Without the demand of training in reality, humans can easily detect a known concept simply based on its language description. Empowering deep learning with this ability undoubtedly enables the neural network to handle complex vision tasks, e.g., object detection, without collecting and annotating real images. To this end, this paper introduces a novel challenging learning paradigm Imaginary-Supervised Object Detection (ISOD), where neither real images nor manual annotations are allowed for training object detectors. To resolve this challenge, we propose ImaginaryNet, a framework to synthesize images by combining pretrained language model and text-to-image synthesis model. Given a class label, the language model is used to generate a full description of a scene with a target object, and the text-to-image model deployed to generate a photo-realistic image. With the synthesized images and class labels, weakly supervised object detection can then be leveraged to accomplish ISOD. By gradually introducing real images and manual annotations, ImaginaryNet can collaborate with other supervision settings to further boost detection performance. Experiments show that ImaginaryNet can (i) obtain about 70% performance in ISOD compared with the weakly supervised counterpart of the same backbone trained on real data, (ii) significantly improve the baseline while achieving state-of-the-art or comparable performance by incorporating ImaginaryNet with other supervision settings.
Masked Autoencoding (MAE) has emerged as an effective approach for pre-training representations across multiple domains. In contrast to discrete tokens in natural languages, the input for image MAE is continuous and subject to additional specifications. We systematically study each input specification during the pre-training stage, and find sequence length is a key axis that further scales MAE. Our study leads to a long-sequence version of MAE with minimal changes to the original recipe, by just decoupling the mask size from the patch size. For object detection and semantic segmentation, our long-sequence MAE shows consistent gains across all the experimental setups without extra computation cost during the transfer. While long-sequence pre-training is discerned most beneficial for detection and segmentation, we also achieve strong results on ImageNet-1K classification by keeping a standard image size and only increasing the sequence length. We hope our findings can provide new insights and avenues for scaling in computer vision.
In recent years, convolutional neural networks (CNNs) are used in a large number of tasks in computer vision. One of them is object detection for autonomous driving. Although CNNs are used widely in many areas, what happens inside the network is still unexplained on many levels. Our goal is to determine the effect of Intrinsic dimension (i.e. minimum number of parameters required to represent data) in different layers on the accuracy of object detection network for augmented data sets. Our investigation determines that there is difference between the representation of normal and augmented data during feature extraction.
Oriented object detection emerges in many applications from aerial images to autonomous driving, while many existing detection benchmarks are annotated with horizontal bounding box only which is also less costive than fine-grained rotated box, leading to a gap between the readily available training corpus and the rising demand for oriented object detection. This paper proposes a simple yet effective oriented object detection approach called H2RBox merely using horizontal box annotation for weakly-supervised training, which closes the above gap and shows competitive performance even against those trained with rotated boxes. The cores of our method are weakly- and self-supervised learning, which predicts the angle of the object by learning the consistency of two different views. To our best knowledge, H2RBox is the first horizontal box annotation-based oriented object detector. Compared to an alternative i.e. horizontal box-supervised instance segmentation with our post adaption to oriented object detection, our approach is not susceptible to the prediction quality of mask and can perform more robustly in complex scenes containing a large number of dense objects and outliers. Experimental results show that H2RBox has significant performance and speed advantages over horizontal box-supervised instance segmentation methods, as well as lower memory requirements. While compared to rotated box-supervised oriented object detectors, our method shows very close performance and speed. The source code is available at PyTorch-based \href{https://github.com/yangxue0827/h2rbox-mmrotate}{MMRotate} and Jittor-based \href{https://github.com/yangxue0827/h2rbox-jittor}{JDet}.
Recent research about camouflaged object detection (COD) aims to segment highly concealed objects hidden in complex surroundings. The tiny, fuzzy camouflaged objects result in visually indistinguishable properties. However, current single-view COD detectors are sensitive to background distractors. Therefore, blurred boundaries and variable shapes of the camouflaged objects are challenging to be fully captured with a single-view detector. To overcome these obstacles, we propose a behavior-inspired framework, called Multi-view Feature Fusion Network (MFFN), which mimics the human behaviors of finding indistinct objects in images, i.e., observing from multiple angles, distances, perspectives. Specifically, the key idea behind it is to generate multiple ways of observation (multi-view) by data augmentation and apply them as inputs. MFFN captures critical boundary and semantic information by comparing and fusing extracted multi-view features. In addition, our MFFN exploits the dependence and interaction between views and channels. Specifically, our methods leverage the complementary information between different views through a two-stage attention module called Co-attention of Multi-view (CAMV). And we design a local-overall module called Channel Fusion Unit (CFU) to explore the channel-wise contextual clues of diverse feature maps in an iterative manner. The experiment results show that our method performs favorably against existing state-of-the-art methods via training with the same data. The code will be available at https://github.com/dwardzheng/MFFN_COD.
Spatial-wise dynamic convolution has become a promising approach to improving the inference efficiency of deep networks. By allocating more computation to the most informative pixels, such an adaptive inference paradigm reduces the spatial redundancy in image features and saves a considerable amount of unnecessary computation. However, the theoretical efficiency achieved by previous methods can hardly translate into a realistic speedup, especially on the multi-core processors (e.g. GPUs). The key challenge is that the existing literature has only focused on designing algorithms with minimal computation, ignoring the fact that the practical latency can also be influenced by scheduling strategies and hardware properties. To bridge the gap between theoretical computation and practical efficiency, we propose a latency-aware spatial-wise dynamic network (LASNet), which performs coarse-grained spatially adaptive inference under the guidance of a novel latency prediction model. The latency prediction model can efficiently estimate the inference latency of dynamic networks by simultaneously considering algorithms, scheduling strategies, and hardware properties. We use the latency predictor to guide both the algorithm design and the scheduling optimization on various hardware platforms. Experiments on image classification, object detection and instance segmentation demonstrate that the proposed framework significantly improves the practical inference efficiency of deep networks. For example, the average latency of a ResNet-101 on the ImageNet validation set could be reduced by 36% and 46% on a server GPU (Nvidia Tesla-V100) and an edge device (Nvidia Jetson TX2 GPU) respectively without sacrificing the accuracy. Code is available at https://github.com/LeapLabTHU/LASNet.