The Hidden Cost of Poor Annotations: How Label Quality Affects Camouflaged Object Detection Performance

Object detection under imperfect data receives great attention recently. Weakly supervised object detection (WSOD) suffers from severe localization issues due to the lack of instance-level annotation, while semi-supervised object detection (SSOD) remains challenging led by the inter-image discrepancy between labeled and unlabeled data. In this study, we propose the Single Instance annotated Object Detection (SIOD), requiring only one instance annotation for each existing category in an image. Degraded from inter-task (WSOD) or inter-image (SSOD) discrepancies to the intra-image discrepancy, SIOD provides more reliable and rich prior knowledge for mining the rest of unlabeled instances and trades off the annotation cost and performance. Under the SIOD setting, we propose a simple yet effective framework, termed Dual-Mining (DMiner), which consists of a Similarity-based Pseudo Label Generating module (SPLG) and a Pixel-level Group Contrastive Learning module (PGCL). SPLG firstly mines latent instances from feature representation space to alleviate the annotation missing problem. To avoid being misled by inaccurate pseudo labels, we propose PGCL to boost the tolerance to false pseudo labels. Extensive experiments on MS COCO verify the feasibility of the SIOD setting and the superiority of the proposed method, which obtains consistent and significant improvements compared to baseline methods and achieves comparable results with fully supervised object detection (FSOD) methods with only 40% instances annotated. Code is available at https://github.com/solicucu/SIOD.

Object detection models have become the current tool of choice for plant disease detection in precision agriculture. Most existing research improved the performance by ameliorating networks and optimizing the loss function. However, because of the vast influence of data annotation quality and the cost of annotation, the data-centric part of a project also needs more investigation. We should further consider the relationship between data annotation strategies, annotation quality, and the model’s performance. In this paper, a systematic strategy with four annotation strategies for plant disease detection is proposed: local, semi-global, global, and symptom-adaptive annotation. Labels with different annotation strategies will result in distinct models’ performance, and their contrasts are remarkable. An interpretability study of the annotation strategy is conducted by using class activation maps. In addition, we define five types of inconsistencies in the annotation process and investigate the severity of the impact of inconsistent labels on model’s performance. Finally, we discuss the problem of label inconsistency during data augmentation. Overall, this data-centric quantitative analysis helps us to understand the significance of annotation strategies, which provides practitioners a way to obtain higher performance and reduce annotation costs on plant disease detection. Our work encourages researchers to pay more attention to annotation consistency and the essential issues of annotation strategy. The code will be released at: https://github.com/JiuqingDong/PlantDiseaseDetection_Yolov5 .

Efficient object detection based on selective attention

Moving object detection and tracking have various applications, including surveillance, anomaly detection, vehicle navigation, etc. The literature on object detection and tracking is rich enough, and there exist several essential survey papers. However, the research on camouflage object detection and tracking is limited due to the complexity of the problem. Existing work on this problem has been done based on either biological characteristics of the camouflaged objects or computer vision techniques. In this paper, we review the existing camouflaged object detection and tracking techniques using computer vision algorithms from the theoretical point of view. This paper also addresses several issues of interest as well as future research direction in this area. We hope this paper will help the reader to learn the recent advances in camouflaged object detection and tracking.

A systematic review of image-level camouflaged object detection with deep learning

Surface defect detection on steel strips is a critical step in quality control for industrial products. While existing research has made some progress in optimizing annotation strategies and improving efficiency, issues such as feature aliasing during the annotation process, the insufficient utilization of boundary information, and the inaccurate representation of complex defect patterns remain inadequately addressed. To tackle these challenges, this paper proposes an annotation optimization framework from the perspective of feature analysis. The framework decomposes defect features into geometric features and grayscale distribution features and, based on feature decoupling theory, classifies defects into three typical patterns: block, linear, and textured defects. For each pattern, the minimum annotation units that preserved essential features were designed, enabling the standardized representation of complex defects and precise boundary localization. Experiments on the NEU-DET dataset showed that this annotation framework improves the average mAP of six mainstream detection models by 4.9 percentage points, validating its effectiveness in enhancing the detection performance. Additionally, this paper introduces an Efficiency–Cost Ratio (ECR) evaluation metric to quantify the relationship between the annotation cost and performance improvement. The study found that block and linear defect detection achieved optimal performance with only 50% annotation effort. This research not only improved the performance of defect detection models but also quantified the annotation resource utilization efficiency, providing robust theoretical support and practical guidance for efficient defect detection in complex industrial scenarios.

Camouflaged object detection (COD), in a broad sense, aims to detect image objects that have high degrees of similarity to the background. COD is more challenging than conventional object detection because of the high degree of “fusion” between a camouflaged object and the background. In this paper, we focused on the accurate detection of camouflaged objects, conducting an in-depth study on COD and addressing the common detection problems of high miss rates and low confidence levels. We proposed a ternary cascade perception-based method for detecting camouflaged objects and constructed a cascade perception network (CPNet). The innovation lies in the proposed ternary cascade perception module (TCPM), which focuses on extracting the relationship information between features and the spatial information of the camouflaged target and the location information of key points. In addition, a cascade aggregation pyramid (CAP) and a joint loss function have been proposed to recognize camouflaged objects accurately. We conducted comprehensive experiments on the COD10K dataset and compared our proposed approach with other seventeen-object detection models. The experimental results showed that CPNet achieves optimal results in terms of six evaluation metrics, including an average precision (AP)50 that reaches 91.41, an AP75 that improves to 73.04, and significantly higher detection accuracy and confidence.

Object detection is a task that performs position identification and label classification of objects in images or videos. The information obtained through this process plays an essential role in various tasks in the field of computer vision. In object detection, the data utilized for training and validation typically originate from public datasets that are well-balanced in terms of the number of objects ascribed to each class in an image. However, in real-world scenarios, handling datasets with much greater class imbalance, i.e., very different numbers of objects for each class, is much more common, and this imbalance may reduce the performance of object detection when predicting unseen test images. In our study, thus, we propose a method that evenly distributes the classes in an image for training and validation, solving the class imbalance problem in object detection. Our proposed method aims to maintain a uniform class distribution through multi-label stratification. We tested our proposed method not only on public datasets that typically exhibit balanced class distribution but also on private datasets that may have imbalanced class distribution. We found that our proposed method was more effective on datasets containing severe imbalance and less data. Our findings indicate that the proposed method can be effectively used on datasets with substantially imbalanced class distribution.

Object detection is an essential function for mobile robots, allowing them to carry out missions efficiently. In recent years, various deep learning models based on convolutional neural networks have achieved good performance in object detection. However, in cases in which robots have to carry out missions in a particular environment, utilizing a model that has been trained without considering the environment in which robots must conduct their tasks degrades their object detection performance, leading to failed missions. This poor model accuracy occurs because of the class imbalance problem, in which the occurrence frequencies of the object classes in the training dataset are significantly different. In this study, we propose a systematic solution that can solve the class imbalance problem by training multiple object detection models and using these models effectively for robots that move through various environments to carry out missions. Moreover, we show through experiments that the proposed multi-model-based object detection framework with environment-context awareness can effectively overcome the class imbalance problem. As a result of the experiment, CPU usage decreased by 45.49% and latency decreased by more than 60%, while object detection accuracy increased by 6.6% on average.

Tactical reconnaissance using small unmanned aerial vehicles has become a common military scenario. However, since their sensor systems are usually limited to rudimentary visual or thermal imaging, the detection of camouflaged objects can be a particularly hard challenge. With respect to SWaP-C criteria, multispectral sensors represent a promising solution to increase the spectral information that could lead to unveiling camouflage. Therefore, this paper investigates and evaluates the applicability of four well-known hyperspectral anomaly detection methods (RX, LRX, CRD, and AED) and a method developed by the authors called local point density (LPD) for near real-time camouflage detection in multispectral imagery based on a specially created dataset. Results show that all targets in the dataset could successfully be detected with an AUC greater than 0.9 by multiple methods, with some methods even reaching an AUC relatively close to 1.0 for certain targets. Yet, great variations in detection performance over all targets and methods were observed. The dataset was additionally enhanced by multiple vegetation indices (BNDVI, GNDVI, and NDRE), which resulted in generally higher detection performances of all methods. Overall, the results demonstrated the general applicability of the hyperspectral anomaly detection methods for camouflage detection in multispectral imagery.

Recently, 3D indoor object detection has become increasingly important in applications such as service robotics, augmented reality and smart homes. However, due to factors like densely arranged objects, severe occlusion and sparse point cloud data, existing methods often suffer from complex architectures, high computational costs or poor performance in small object detection, making it difficult to balance accuracy and efficiency. To address these challenges, we propose an efficient multi‐scale feature fusion model for 3D indoor object detection (EMF3D), offering an end‐to‐end, lightweight solution tailored for indoor point cloud scenarios. The method employs sparse voxel convolution to build a compact feature extraction network and introduces a sparse convolution‐based squeeze‐and‐excitation block at the feature fusion stage to adaptively learn feature channel weights. Furthermore, the enhanced fusion module (EFM) strengthens the perception of critical structures and improves feature discriminability by aggregating multi‐scale representations from earlier attention stages. Extensive experiments are conducted on three indoor point cloud datasets–ScanNet V2, SUN RGB‐D and S3DIS. Results show that the proposed method outperforms major detection approaches across multiple metrics. Compared with existing methods, EMF3D achieves superior performance in small object detection and offers a better trade‐off between accuracy and efficiency.

With the wide applications of remote sensing technology in engineering, the demands of efficient object detection algorithms for remote sensing images have also been significantly increased in recent years. Traditional detection methods have the shortcomings of low accuracy and poor robustness, which is difficult to be applied to remote sensing images with complex background and varieties of objects. Recently, the deep convolutional neural networks have already shown great advances in object detection and outperformed many traditional methods. In this work, we study the performance of a region proposal-based method, Mask R-CNN, for detecting airplane and ship in remote sensing images. Specifically, we add an FPN module to improve the accuracy of small objects detection, and a mask branch is used to describe the shape of objects. In addition, we used a series of data augmentation strategies during training for meeting the CNN’s requirement of training samples diversity. The experimental results show that our model has superior performance in object detection of remote sensing images.

Underwater image enhancement is often perceived as a disadvantageous process to object detection. We propose a novel analysis of the interactions between enhancement and detection, elaborating on the potential of enhancement to improve detection. In particular, we evaluate object detection performance for each individual image rather than across the entire set to allow a direct performance comparison of each image before and after enhancement. This approach enables the generation of unique queries to identify the outperforming and underperforming enhanced images compared to the original images. To accomplish this, we first produce enhanced image sets of the original images using recent image enhancement models. Each enhanced set is then divided into two groups: (1) images that outperform or match the performance of the original images and (2) images that underperform. Subsequently, we create mixed original-enhanced sets by replacing underperforming enhanced images with their corresponding original images. Next, we conduct a detailed analysis by evaluating all generated groups for quality and detection performance attributes. Finally, we perform an overlap analysis between the generated enhanced sets to identify cases where the enhanced images of different enhancement algorithms unanimously outperform, equally perform, or underperform the original images. Our analysis reveals that, when evaluated individually, most enhanced images achieve equal or superior performance compared to their original counterparts. The proposed method uncovers variations in detection performance that are not apparent in a whole set as opposed to a per-image evaluation because the latter reveals that only a small percentage of enhanced images cause an overall negative impact on detection. We also find that over-enhancement may lead to deteriorated object detection performance. Lastly, we note that enhanced images reveal hidden objects that were not annotated due to the low visibility of the original images.

Fire detection from remote sensing images plays a critical role in early fire warning, fire risk reduction, and post-fire damage assessment. Though deep learning-based approaches have achieved huge successes in fire detection and locating, their data-driven nature requires substantial human resources and time for meticulously labeling pixel-level annotations. The challenge of limited well-annotated labels is still hampering its application. In this paper, we propose a novel label evolution method, Box2Pix, that is to automatically transform box-level labels into pixel-levels labels, alleviating the heavy labor and time cost of pixel-wise fire annotation. Box2Pix leverages box-level annotated object detection datasets with the pre-trained large segmentation models to generate pixel-level annotated semantic segmentation datasets to further boost fire detection performance. Moreover, we introduce innovative point prompt generation methods by leveraging superpixel segmentation and saliency detection to identify potential target regions for generating random points, augmenting the semantic segmentation capability of Segment Anything Model (SAM). Box2Pix eliminates the need for extensive human involvement, avoids reliance on large-scale datasets for training or fine-tuning, enhancing the speed and quality of pixel-level label generation while reducing annotation difficulty and costs. The experiment on the highly generalized Flame And Smoke Detection Dataset (FASDD) demonstrates its superiority over the existing state-of-the-art methods, with improvements of 52.51% in mIoU and 41.38% in overall accuracy. Based on Box2Pix, the largest pixel-level fire detection dataset, Flame And Smoke Segmentation Dataset (FASSD) is generated. The newly created dataset is accessible on: https://github.com/OyamingO/Box2Pix.

Most Camouflaged Object Detection (COD) methods heavily rely on mask annotations, which are time-consuming and labor-intensive to acquire. Existing weakly-supervised COD approaches exhibit significantly inferior performance compared to fully-supervised methods and struggle to simultaneously support all the existing types of camouflaged object labels, including scribbles, bounding boxes, and points. Even for Segment Anything Model (SAM), it is still problematic to handle the weakly-supervised COD and it typically encounters challenges of prompt compatibility of the scribble labels, extreme response, semantically erroneous response, and unstable feature representations, producing unsatisfactory results in camouflaged scenes. To mitigate these issues, we propose a unified COD framework in this paper, termed SAM-COD, which is capable of supporting arbitrary weakly-supervised labels. Our SAM-COD employs a prompt adapter to handle scribbles as prompts based on SAM. Meanwhile, we introduce response filter and semantic matcher modules to improve the quality of the masks obtained by SAM under COD prompts. To alleviate the negative impacts of inaccurate mask predictions, a new strategy of prompt-adaptive knowledge distillation is utilized to ensure a reliable feature representation. To validate the effectiveness of our approach, we have conducted extensive empirical experiments on three mainstream COD benchmarks. The results demonstrate the superiority of our method against state-of-the-art weakly-supervised and even fully-supervised methods.