Heatmap-based Explanation of YOLOv5 Object Detection with Layer-wise Relevance Propagation

Karasmanoglou, Apostolos; Antonakakis, Marios; Zervakis, Michalis

doi:10.1109/ist55454.2022.9827744

Cited by 13 publications

(7 citation statements)

References 28 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…We used ground truth bounding boxes to calculate heatmaps using LRP. We use [4] to calculate heatmaps using ground-truth bounding boxes. After calculating heatmaps, based on active pixels contributing to ground-truth bounding boxes, we calculate the histogram of intensities of active pixels whose intensities are normalized using the sum of intensities of all active pixels.…”

Section: Evaluation Methods Using Lrpmentioning

confidence: 99%

Quantized deep learning for low-cost IR sensors with layer-wise relevance propagation

Miwa,

Susumu

2023

Infrared Technology and Applications XLIX

View full text Add to dashboard Cite

In recent years, the market for infrared (IR) sensors has expanded from traditional defense and security applications to include consumer products. As a result, there is increasing demand for embedded IR systems that integrate low-cost sensors with embedded processors. Meanwhile, deep learning has made significant advances and achieved superhuman performance in some domains. To support deep learning in embedded systems, new processors have emerged that are specifically designed for running deep neural networks. In this paper, we propose a performance evaluation method for applying quantized deep learning to low-cost IR sensors using layer-wise relevance propagation. Our method provides visualized analysis of what the neural networks learn. We demonstrate the effectiveness of our approach through experiments on a low-cost IR sensor dataset, showing that our method achieves an explainable performance evaluation method for degraded cases arising from the tradeoff between speed and accuracy of quantized detectors which is a typical problem in embedded systems with limited computational resources.

show abstract

Section: Evaluation Methods Using Lrpmentioning

confidence: 99%

Quantized deep learning for low-cost IR sensors with layer-wise relevance propagation

Miwa,

Susumu

2023

Infrared Technology and Applications XLIX

View full text Add to dashboard Cite

show abstract

“…In XAI for object detection, many methods extend the existing methods designed for image classification by considering both localization and classification of a target object detection. As an application of back-propagation-based methods, Karasmanoglou et al [19] applied CRP to the YOLO detectors [20,21], which are widely utilized due to their high processing speed, to explain detection results. In this method, the calculation of the relevances is restricted to regions near the target bounding box, aligning with the target class label.…”

Section: A Visualization Of Explanationmentioning

confidence: 99%

“…1 shows the comparison of these methods, with the detection results from the small YOLOv5 model (YOLOv5s) 1 set as an explanation target. Back-propagationbased [19] and activation-map-based methods [5] offer faster processing speed due to their ability to leverage internal information within the object detector. However, these saliency maps may fail to correctly capture the target object.…”

Section: A Visualization Of Explanationmentioning

confidence: 99%

Fast Explanation Using Shapley Value for Object Detection

Kuroki,

Yamasaki

2024

IEEE Access

View full text Add to dashboard Cite

In explainable artificial intelligence (XAI) for object detection, saliency maps are employed to highlight important regions for a learned model's prediction. However, there is a trade-off where the more accurate the explanation results, the higher the computational cost becomes, posing a challenge for practical applications. Therefore, this paper proposes a novel XAI method for object detection that addresses this challenge. In recent years, research on XAI that satisfies desirable properties for explanatory validity by introducing the Shapley value has been widely conducted. However, a common drawback across these approaches is the high computational cost, which has hindered broad implementation. Our proposed method utilizes an explainer model that learns to estimate the Shapley value and provides a reliable explanation for object detection in a real-time inference. This framework can be applied to various object detectors in a model-agnostic manner. We experimentally demonstrate in quantitative evaluation that our method achieves the fastest explanation while delivering superior performance compared to other existing methods.

show abstract

“…Some methods extend those for image classification by adding conditions on the calculation to limit an explanation scope to a target object. Contrastive Relevance Propagation (CRP) for the You Only Look Once (YOLO) detector [3] confines the calculation of attributions to those originating from nearby the target bounding box and denoting the class label associated with the target. On the other hand, D-RISE [4] extends the modelagnostic approach [5] and calculates the feature attribution by sampling masked images and their corresponding output scores.…”

Section: Introductionmentioning

confidence: 99%

BSED: Baseline Shapley-Based Explainable Detector

Kuroki,

Yamasaki

2024

IEEE Access

View full text Add to dashboard Cite

Explainable artificial intelligence (XAI) has witnessed significant advances in the field of object recognition, with saliency maps being used to highlight image features relevant to the predictions of learned models. Although these advances have made artificial intelligence (AI)-based technology more interpretable to humans, several issues have come to light, as some approaches present explanations irrelevant to predictions, and cannot guarantee the validity of XAI (axioms). In this study, we propose the Baseline Shapley-based Explainable Detector (BSED), which extends the Shapley value to object detection for images, thereby enhancing the validity of interpretation. The Shapley value can attribute the prediction of a learned model to a baseline feature while satisfying the explainability axioms. The processing cost for the BSED is within the reasonable range, while the original Shapley value is prohibitively computationally expensive. Furthermore, BSED is a generalizable method that can be applied to various object detectors for images in a model-agnostic manner, and interpret various detection targets without fine-grained parameter tuning. These strengths can enable the practical applicability of XAI. We present quantitative and qualitative evaluations to demonstrate that our method outperforms existing methods in terms of explanation validity. Moreover, we present some applications, such as correcting detection based on explanations from our method.

show abstract

Heatmap-based Explanation of YOLOv5 Object Detection with Layer-wise Relevance Propagation

Cited by 13 publications

References 28 publications

Quantized deep learning for low-cost IR sensors with layer-wise relevance propagation

Quantized deep learning for low-cost IR sensors with layer-wise relevance propagation

Fast Explanation Using Shapley Value for Object Detection

BSED: Baseline Shapley-Based Explainable Detector

Contact Info

Product

Resources

About