Drones as collaborative sensors for image recognition

Zientara, Peter A.; Choi, Jinhang; Sampson, Jack; Narayanan, Vijaykrishnan

doi:10.1109/icce.2018.8326187

Cited by 5 publications

(3 citation statements)

References 6 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Finally, it might instead be required to give a partial classification, such as the category of an object e.g., it might recognise an area of land is flooded, but may not be able to infer the type of land that is underneath. (Some examples of vision models in drone systems are given in: Zientara et al (2018), van Gemert et al (2015 , Radovic et al (2017))…”

Section: Defining Graceful Degradationmentioning

confidence: 99%

Graceful Degradation and Related Fields

Dymond

2021

Preprint

View full text Add to dashboard Cite

When machine learning models encounter data which is out of the distribution on which they were trained they have a tendency to behave poorly, most prominently over-confidence in erroneous predictions. Such behaviours will have disastrous effects on real-world machine learning systems. In this field graceful degradation refers to the optimisation of model performance as it encounters this out-of-distribution data. This work presents a definition and discussion of graceful degradation and where it can be applied in deployed visual systems. Following this a survey of relevant areas is undertaken, novelly splitting the graceful degradation problem into active and passive approaches. In passive approaches, graceful degradation is handled and achieved by the model in a self-contained manner, in active approaches the model is updated upon encountering epistemic uncertainties. This work communicates the importance of the problem and aims to prompt the development of machine learning strategies that are aware of graceful degradation.

show abstract

Section: Defining Graceful Degradationmentioning

confidence: 99%

Graceful Degradation and Related Fields

Dymond

2021

Preprint

View full text Add to dashboard Cite

show abstract

“…For the practical SOC dataset, it was collected from NORAD_Catalog (North American Aerospace Defense Command) 1 and UCS_Satellite database (Union of Concerned Scientists). 2 The NORAD_Catalog contains 8071 space object examples, with 9 attributes (cospar_id, nord_id, period, perigee, apogee, eccentricity, rcs, size, amr) and 3 categories (Debris, RocketBody and Satellite). UCS_Satellite database contains 1346 satellite object examples, with 17 attributes (cospar_id, nord_id, period, perigee, Besides, the multi-granular categories of objects is illustrated in Figure 2.…”

Section: Empirical Evaluationmentioning

confidence: 99%

“…As multi-sensor networks for object classification become increasingly complex, it has been one of the most important requirements to build explainable artificial intelligence (XAI) systems for human users [1], [2]. Since most of intelligent algorithms to classify objects are lack of transparency and interpretability (i.e.…”

mentioning

confidence: 99%

Leveraging Cognitive Context Knowledge for Argumentation-Based Object Classification in Multi-Sensor Networks

Hao

Liu

et al. 2019

IEEE Access

View full text Add to dashboard Cite

It is a great challenge to achieve interpretable collaborative object classification in multisensor networks. In this situation, argumentation-based object classification has been considered a promising paradigm, due to its natural means of justifying and explaining complicated decision making within multiple agents. However, disagreements between sensor agents are often encountered because of various object category levels. To address this category of granularity inconsistent problem in multi-sensor collaborative object classification tasks, we propose a cognitive context knowledge-enriched method for classification conflict resolution. The cognitive context is concerned, in this paper, to investigate how rich contextual knowledge-equipped cognitive agents can facilitate semantic consensus in argumentation-based object classification. The empirical evaluation demonstrates the effectiveness of our method with improvement over state-of-the-art, especially in the presence of noisy sensor data, while giving argumentative explanations. Therefore, it is suggested that people who can benefit from the proposed method in this paper are the human user of multi-sensor object classification systems, in which explaining decision support is one of the important factors concerned.

show abstract

Object Recognition Through UAV Observations Based on Yolo and Generative Adversarial Network

Gan

Neretin

et al. 2021

Lecture Notes of the Institute for Computer Sciences, Social Informatics and Telecommunications Engineering

View full text Add to dashboard Cite

Drones as collaborative sensors for image recognition

Cited by 5 publications

References 6 publications

Graceful Degradation and Related Fields

Graceful Degradation and Related Fields

Leveraging Cognitive Context Knowledge for Argumentation-Based Object Classification in Multi-Sensor Networks

Object Recognition Through UAV Observations Based on Yolo and Generative Adversarial Network

Contact Info

Product

Resources

About