Measuring the dynamics of camouflage in natural scenes using convolutional neural networks

Burg, Erik Van der; Hogervorst, Maarten A.; Toet, Alexander

doi:10.1117/12.2636107

“…Since identification conspicuity and the level of confidence predicted by YOLO both increase with the visible amount of characteristic object details we expect that both object distinctness measures will be correlated. A part of this work was previously published as a conference paper [21].…”

Section: Yolo Camouflage Metricmentioning

confidence: 99%

A Convolutional Neural Network as a Potential Tool for Camouflage Assessment

Van der Burg,

Toet,

Perone

et al. 2024

Preprint

0

View full text Add to dashboard Cite

Camouflage evaluation typically involves human visual search and detection experiments that are time-consuming and expensive. Hence, there is a need for models that compute camouflage effectiveness from digital imagery. Convolutional neural networks are a powerful tool for automatic object detection and recognition. We investigated whether such a network (YOLO) can also provide a measure of camouflage effectiveness that is related to human perception. To this end, human performance measures of camouflage effectiveness such as detection time and target conspicuity were obtained in observer experiments and compared with the performance of the (standard, i.e. pre-trained) YOLO-V4-tiny algorithm. YOLO provides the probability that a detected object is correctly classified, and this is adopted as our measure of camouflage effectiveness. We compared the YOLO-predicted classification probability for a person in camouflage clothing moving through rural and urban scenes to human detection performance. Overall, camouflage effectiveness predicted by YOLO correlates with human observer performance. At close distances, YOLO’s classification performance appears sensitive to high-contrast shape-breaking elements of a camouflage pattern. This suggests that YOLO may be adapted to assess camouflage effectiveness for a wide range of applications, such as evaluating or optimizing one’s signature and predicting optimal hiding locations in a given environment. Further research is required to fully establish YOLO’s limitations and applicability for this purpose.

show abstract

“…Since identification conspicuity and the level of confidence predicted by YOLO both increase with the visible amount of characteristic object details we expect that both object distinctness measures will be correlated. A part of this work was previously published as a conference paper [25]. (a) Bush/desert background (Soesterduinen) and urban background (Amsterdam).…”

Section: Yolo Camouflage Metricmentioning

confidence: 99%

A Convolutional Neural Network as a Potential Tool for Camouflage Assessment

Van der Burg,

Toet,

Perone

et al. 2024

Preprint

0

View full text Add to dashboard Cite

Camouflage evaluation typically involves human visual search and detection experiments that are time-consuming and expensive. Hence, there is a need for models that compute camouflage effectiveness from digital imagery. Convolutional neural networks are a powerful tool for automatic object detection and recognition. We investigated whether such a network (YOLO) can also provide a measure of camouflage effectiveness that is related to human perception. To this end, human performance measures of camouflage effectiveness such as detection time and target conspicuity were obtained in observer experiments and compared with the performance of the (standard, i.e. pre-trained) YOLO-V4-tiny algorithm. YOLO provides the probability that a detected object is correctly classified, and this is adopted as our measure of camouflage effectiveness. We compared the YOLO-predicted classification probability for a person in camouflage clothing moving through rural and urban scenes to human detection performance. Overall, camouflage effectiveness predicted by YOLO correlates with human observer performance. At close distances, YOLO’s classification performance appears sensitive to high-contrast shape-breaking elements of a camouflage pattern. This suggests that YOLO may be adapted to assess camouflage effectiveness for a wide range of applications, such as evaluating or optimizing one’s signature and predicting optimal hiding locations in a given environment. Further research is required to fully establish YOLO’s limitations and applicability for this purpose.

show abstract

A Neural Network Model for Detecting Camouflaged Objects

Tsybulia,

Tymchuk,

Tsybulia

2024

Проблеми Створення, Випробування, Застосування Та Експлуатації Складних Інформаційних Систем

0

View full text Add to dashboard Cite

This article discusses the use of computer vision to detect camouflaged objects that are hidden by natural masks and camouflage devices. Such objects are usually varied in size, fuzzy, and visually merge with the terrain, which makes them difficult to detect. The authors analyze the models of machine learning algorithms used to segment and detect objects in images. Taking into account the analysis, to solve the problem of detecting camouflaged objects, the paper proposes a neural network model with an encoder-decoder architecture. Its features are: the use of an additional layer at the input, which is fed with an image processed by a Sobel filter, which allows to enhance the detection of object edges; the use of the convolutional stretching algorithm in the encoder blocks, in parallel with the main part of the key features determination, leads to a decrease in the dependence of detection on the size of objects; the use of a mechanism in the decoder blocks to focus on important parts of the image increases the probability of correct classification of image areas in cases of uncertainty of the model regarding their. Experiments by modeling, with different hyperparameters of the neural network, allowed us to determine that binary cross-entropy is most suitable as a loss function for solving the problem of detecting objects with strong background noise, and the choice of Parametric Rectified Linear Unit as an activation function allows to improve the quality of object segmentation. We also consider the use of various metrics to evaluate the effectiveness of the created model. Testing on datasets with real cloaked objects allowed us to identify problematic issues affecting the segmentation process in general and the accuracy of detecting cloaked objects in particular, the solution of which can improve the efficiency of neural networks in object detection. The results of the research are proposed to be used in the creation of camouflage means to determine their effectiveness, as well as to search for camouflaged enemy objects in the course of intelligence processing.

show abstract

Measuring the dynamics of camouflage in natural scenes using convolutional neural networks

Cited by 3 publications

References 25 publications

A Convolutional Neural Network as a Potential Tool for Camouflage Assessment

A Convolutional Neural Network as a Potential Tool for Camouflage Assessment

A Convolutional Neural Network as a Potential Tool for Camouflage Assessment

A Neural Network Model for Detecting Camouflaged Objects

Contact Info

Product

Resources

About