Leveraging Classification Metrics for Quantitative System-Level Analysis with Temporal Logic Specifications

Badithela, Apurva; Wongpiromsarn, Tichakorn; Murray, Richard M.

doi:10.1109/cdc45484.2021.9683611

Cited by 9 publications

(6 citation statements)

References 20 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…All of the models were trained for 300 epochs and using an Nvidia RTX 3060 GPU. In object detection, precision and recall are essential metrics used to evaluate the performance of a trained model in correctly identifying detected objects [21]. Precision (4) measures the proportion of positive vehicle identifications that were accurately identified using true positive (TP) and false positive (FP) detections.…”

Section: Resultsmentioning

confidence: 99%

“…Precision (4) measures the proportion of positive vehicle identifications that were accurately identified using true positive (TP) and false positive (FP) detections. On the other hand, recall (5) measures the proportion of actual positive identifications that were correctly identified using true positive (TP) and false negative (FN) detections [21]. A high value of precision and recall indicates that the model can accurately detect all positive vehicles correctly and classify them accurately.…”

Section: Resultsmentioning

confidence: 99%

See 1 more Smart Citation

Real-time vehicle counting using custom YOLOv8n and DeepSORT for resource-limited edge devices

Saadeldin,

Rashid,

Shafie

et al. 2024

TELKOMNIKA

View full text Add to dashboard Cite

Recently, there has been a significant increase in the use of deep learning and low-computing edge devices for analysis of video-based systems, particularly in the field of intelligent transportation systems (ITS). One promising application of computer vision techniques in ITS is in the development of lowcomputing and accurate vehicle counting systems that can be used to eliminate dependence on external cloud computing resources. This paper proposes a compact, reliable and real-time vehicle counting solution which can be deployed on low-computational requirement edge computing devices. The system makes use of a custom-built vehicle detection algorithm based on the you only look once version 8 nano (YOLOv8n), combined with a deep association metric (DeepSORT) object tracking algorithm and an efficient vehicle counting method for accurate counting of vehicles in highway scenes. The system is trained to detect, track and count four distinct vehicle classeses, namely: car, motorcycle, bus, and truck. The proposed system was able to achieve an average vehicle detection mean average precision (mAP) score of 97.5%, a vehicle counting accuracy score of 96.8% and an average speed of 19.4 frames per second (FPS), all while being deployed on a compact Nvidia Jetson Nano edge-computing device. The proposed system outperforms other previously proposed tools in terms of both accuracy and speed.

show abstract

Section: Resultsmentioning

confidence: 99%

Section: Resultsmentioning

confidence: 99%

Real-time vehicle counting using custom YOLOv8n and DeepSORT for resource-limited edge devices

Saadeldin,

Rashid,

Shafie

et al. 2024

TELKOMNIKA

View full text Add to dashboard Cite

show abstract

“…In evaluating object detection methods, many essential ways are used to analyze model performance [17], such as recall accuracy precision IoU and f1 score.…”

Section: Evaluation Methodsmentioning

confidence: 99%

A Fine-Tuned RetinaNet for Real-Time Lettuce Detection

Prasetyo,

Nambo

2024

LKJITI

View full text Add to dashboard Cite

The agricultural industry plays a vital role in the global demand for food production. Along with population growth, there is an increasing need for efficient farming practices that can maximize crop yields. Conventional methods of harvesting lettuce often rely on manual labor, which can be time-consuming, labor-intensive, and prone to human error. These challenges lead to research into automation technology, such as robotics, to improve harvest efficiency and reduce reliance on human intervention. Deep learning-based object detection models have shown impressive success in various computer vision tasks, such as object recognition. RetinaNet model can be trained to identify and localize lettuce accurately. However, the pre-trained models must be fine-tuned to adapt to the specific characteristics of lettuce, such as shape, size, and occlusion, to deploy object recognition models in real-world agricultural scenarios. Fine-tuning the models using lettuce-specific datasets can improve their accuracy and robustness for detecting and localizing lettuce. The data acquired for RetinaNet has the highest accuracy of 0.782, recall of 0.844, f1-score of 0.875, and mAP of 0,962. Metrics evaluate that the higher the score, the better the model performs.

show abstract

“…Building upon the results of [25], the key contributions of this paper are (i) framing the problem of synthesizing test environments for reach-avoid specifications in linear temporal logic (LTL) as a multi-commodity network flow problem, (ii) presenting an efficiently solvable convex-concave min-max optimization-based relaxation that results in a constrained test, (iii) demonstrating the approach by executing the resulting test strategy to reactively test dynamic locomotion behaviors of the Unitree A1 quadruped. A key advantage of our method is that the synthesized test is reactive -the constraints visible to the system under test are reactive to the system state and depend on the system's strategy, which is not known to the tester a priori.…”

Section: Introductionmentioning

confidence: 97%

Synthesizing Reactive Test Environments for Autonomous Systems: Testing Reach-Avoid Specifications with Multi-Commodity Flows

Apurva

Graebener

Mazumdar

et al. 2023

2023 IEEE International Conference on Robotics and Automation (ICRA)

View full text Add to dashboard Cite

We study automated test generation for verifying discrete decision-making modules in autonomous systems. We utilize linear temporal logic to encode the requirements on the system under test in the system specification and the behavior that we want to observe during the test is given as the test specification which is unknown to the system. First, we use the specifications and their corresponding non-deterministic B üchi automata to generate the specification product automaton. Second, a virtual product graph representing the high-level interaction between the system and the test environment is constructed modeling the product automaton encoding the system, the test environment, and specifications. The main result of this paper is an optimization problem, framed as a multicommodity network flow problem, that solves for constraints on the virtual product graph which can then be projected to the test environment. Therefore, the result of the optimization problem is reactive test synthesis that ensures that the system meets the test specifications along with satisfying the system specifications. This framework is illustrated in simulation on grid world examples, and demonstrated on hardware with the Unitree A1 quadruped, wherein dynamic locomotion behaviors are verified in the context of reactive test environments.

show abstract

Leveraging Classification Metrics for Quantitative System-Level Analysis with Temporal Logic Specifications

Cited by 9 publications

References 20 publications

Real-time vehicle counting using custom YOLOv8n and DeepSORT for resource-limited edge devices

Real-time vehicle counting using custom YOLOv8n and DeepSORT for resource-limited edge devices

A Fine-Tuned RetinaNet for Real-Time Lettuce Detection

Synthesizing Reactive Test Environments for Autonomous Systems: Testing Reach-Avoid Specifications with Multi-Commodity Flows

Contact Info

Product

Resources

About