Key strategies for synthetic data generation for training intelligent systems based on people detection from omnidirectional cameras

Aranjuelo, Nerea; García, Sergio Castro; Loyo, Estíbaliz; Unzueta, Luis; Otaegui, Oihana

doi:10.1016/j.compeleceng.2021.107105

Cited by 16 publications

(9 citation statements)

References 4 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The most basic method is to first train on the synthetic domain and then fine-tune the model on the real domain [29]. An alternative method is to jointly train with both domains' images by using mini-batches from source and target domains [30,31]. These methods require having an adequate number of annotated real samples for optimal accuracy.…”

Section: Dnn Training With Synthetic and Real Datamentioning

confidence: 99%

“…The model achieves an accuracy of 93.8%, which is 1.5% lower than our proposed methodology. We additionally compare our method with [30,31], which jointly train with both domains' images by using mini-batches from the source and target domain. We follow the same data strategy and train the classification DNN with no domain adversarial branch.…”

Section: Leveraging Synthetic Data For Dnn-based Visual Analysis Of P...mentioning

confidence: 99%

“…Test Accuracy Pretrain with synthetic data, fine-tune with real data [10,29] 93.8% Train with real and synthetic mini-batches [30,31] 92.2% Train with a domain adversarial strategy (ours) 95.3% Table 7 Comparison between achieved DNN accuracy in the test set for different data combination strategies.…”

Section: Synthetic and Real Data Combination Strategymentioning

confidence: 99%

See 2 more Smart Citations

Leveraging Synthetic Data for DNN-Based Visual Analysis of Passenger Seats

et al. 2022

Self Cite

View full text Add to dashboard Cite

Deep Neural Network (DNN)-based vision systems could improve passenger transportation safety by automating processes such as verifying the correct positioning of luggage, seat occupancy, etc. Abundant and well-distributed data are essential to make DNNs learn appropriate pattern recognition features and have enough generalization ability. The use of synthetic data can reduce the effort of generating varied and annotated data. However, synthetic data usually present a domain gap with real-world samples, that can be reduced with domain adaptation techniques. This paper proposes a methodology to build simulated environments to generate balanced and varied synthetic data and avoid including redundant samples to train classification DNNs for passenger seat analysis. We show a practical implementation for detecting whether luggage is correctly placed or not in an aircraft cabin. Experimental results show the contribution of the synthetic samples and the importance of correctly discarding redundant data.

show abstract

Section: Dnn Training With Synthetic and Real Datamentioning

confidence: 99%

Section: Leveraging Synthetic Data For Dnn-based Visual Analysis Of P...mentioning

confidence: 99%

See 1 more Smart Citation

Leveraging Synthetic Data for DNN-Based Visual Analysis of Passenger Seats

et al. 2022

Self Cite

View full text Add to dashboard Cite

show abstract

“…• People detection/counting denotes the detection or counting of individuals within a given image or video [39] [40].…”

Section: Human Analysismentioning

confidence: 99%

Synthetic Data in Human Analysis: A Survey

Joshi,

Grimmer,

Rathgeb

et al. 2024

IEEE Trans. Pattern Anal. Mach. Intell.

View full text Add to dashboard Cite

Deep neural networks have become prevalent in human analysis, boosting the performance of applications, such as biometric recognition, action recognition, as well as person re-identification. However, the performance of such networks scales with the available training data. In human analysis, the demand for large-scale datasets poses a severe challenge, as data collection is tedious, time-expensive, costly and must comply with data protection laws. Current research investigates the generation of synthetic data as an efficient and privacy-ensuring alternative to collecting real data in the field. This survey introduces the basic definitions and methodologies, essential when generating and employing synthetic data for human analysis. We summarise current state-of-the-art methods and the main benefits of using synthetic data. We also provide an overview of publicly available synthetic datasets and generation models. Finally, we discuss limitations, as well as open research problems in this field. This survey is intended for researchers and practitioners in the field of human analysis.

show abstract

“…A complex nested model has been utilized to create terabytes of structurally similar data for Internet of Things (IoT) research by the authors in [11]. The application of synthetic data generation to create training datasets can be observed in many other fields such as in plasma current quench studies [12], to analyze and predict seismic activities [13], to correctly identify and detect people using omnidirectional cameras [14], wastewater treatment modeling studies [15], and meteorological studies [16,17]. The synthetic data has been used to test the DGA toolbox.…”

Section: Introductionmentioning

confidence: 99%

Generation of synthetic datasets for transformer’s dissolved gas analysis using Monte-Carlo Simulation

Babu

Bashir

Pillai

et al. 2021

2021 56th International Universities Power Engineering Conference (UPEC)

View full text Add to dashboard Cite

The fault diagnosis in power transformers is carried out using Dissolved Gas Analysis (DGA). Although DGA does provide key information for fault detection, the method is inherently complex. Several methods have been developed for DGA, but still possess challenges in accurately detecting the fault. A method has been developed to generate synthetic data using Monte-Carlo simulation. The generated synthetic data is feed into DGA excel tool to investigate the accuracy of fault detection. The synthetic data can be used to further enhance the DGA tool, improve its accuracy and investigate the inclusive faults. A model has been proposed for the integration of synthetic data generator with DGA tool for machine learning and to obtain an automated and improved DGA tool for fault diagnoses in power transformers.

show abstract

Key strategies for synthetic data generation for training intelligent systems based on people detection from omnidirectional cameras

Cited by 16 publications

References 4 publications

Leveraging Synthetic Data for DNN-Based Visual Analysis of Passenger Seats

Leveraging Synthetic Data for DNN-Based Visual Analysis of Passenger Seats

Synthetic Data in Human Analysis: A Survey

Generation of synthetic datasets for transformer’s dissolved gas analysis using Monte-Carlo Simulation

Contact Info

Product

Resources

About