Methodological Selection of Optimal Features for Object Classification Based on Stereovision System
Abstract
:1. Introduction
2. Background Knowledge and Related Works
2.1. Flying Object Features
- 1.
- 2.
- Morphology: The analysis of the size of a bird, its shape (beak, wings, etc.), or plumage colour based on the video or image recordings of optical sensors [16].
- 3.
- Motion and behavioural patterns: The analysis of a bird’s (or other object’s) flight trajectories and their features, including turning angle, curvature, velocity, acceleration, or periodic oscillation due to wing flapping. The data can be gathered using various types of sensors: optical [17], radar [18], or accelerometer [19].
2.2. Optimal Feature Selection Methods
2.3. Flying Object Classifiers
3. Problem Statement, Objectives, and Main Contributions
4. Data Preparation and Feature Preselection
4.1. Data Gathering and Preprocessing
4.2. Features Extraction and Normalisation
- 1.
- Polar angle, , of object localisation in spherical co-ordinates, in radian.
- 2.
- Object angular velocity, calculated for the polar and azimuth angles, and , respectively, in radian/second.
- 3.
- Object size extracted from the image, calculated as the number of an object’s pixels, , in pixels.
- 4.
- Arc path lengths, calculated for polar and azimuth angles, and , respectively, as the path lengths in the polar and azimuth spherical co-ordinates, in meters.
- 1.
- The histograms of the angular velocities and depict the characteristics of the detected object’s flight. The histograms use nine bins for 〈−0.01, 0.01〉 and values; see Figure 3a,b.
- 2.
- The bar diagram of average angular velocities for distance intervals. The features depict changes in flight characteristics with distance, e.g., angular velocities are lower at large distances; see Figure 3c.
- 3.
- The bar diagram of average polar angle, , with respect to distance intervals; see Figure 3f.
- 4.
- The bar diagram of average size, , concerning distance intervals, which strongly depend on the distance, grows as the distance decreases; see Figure 3e.
- 5.
- The bar diagram of average arc path lengths, , and ; see Figure 3e.
- 6.
- The variances of each physical quantity were computed to depict their statistical characteristics.
5. Optimal Feature Selection for Classification
5.1. Ga- and CCF-Based Optimal Feature Selection Methodology
- 1.
- Select n features with the highest absolute correlation coefficient to the class;
- 2.
- Use the selected features to train and evaluate a classifier five times with differently split training and test sets;
- 3.
- Return the average accuracy and recall on the test set from the five iterations.
- Set minimal thresholds and for the accuracy and recall, respectively. These thresholds define our goal performance. For the Aeroplane/Bird classification, they are set to 95 and 97, respectively, and for Big_Raptor/Other_Bird classification, they are 75 and 90, respectively, as is stated in Section 3.
- Use the thresholds to define the two variables: modified accuracy , and modified recall .
- Calculate the fitness value from the following:
5.2. Validation of the Optimal Feature Selection for Different Classifiers
- RF () for Aeroplane/Bird classification;
- RF () for Big_Raptor/Other_Bird Classification.
6. Discussion
Author Contributions
Funding
Institutional Review Board Statement
Informed Consent Statement
Data Availability Statement
Conflicts of Interest
References
- Duriez, O.; Pilard, P.; Saulnier, N.; Boudarel, P.; Besnard, A. Windfarm collisions in medium-sized raptors: Even increasing populations can suffer strong demographic impacts. Anim. Conserv. 2023, 26, 264–275. [Google Scholar] [CrossRef]
- Salkanović, E. Protecting avian wildlife for wind farm siting: The Screening Tool Proof of Concept. Energy Sustain. Dev. 2023, 74, 66–78. [Google Scholar] [CrossRef]
- Bauer, S.; Shamoun-Baranes, J.; Nilsson, C.; Farnsworth, A.; Kelly, J.F.; Reynolds, D.R.; Dokter, A.M.; Krauel, J.F.; Petterson, L.B.; Horton, K.G.; et al. The grand challenges of migration ecology that radar aeroecology can help answer. Ecography 2019, 42, 861–875. [Google Scholar] [CrossRef]
- José Carlos Ribeiro, D.D. Analysis and evaluation of the risk of bird strikes in the international guarulohs airport surroundings. Indep. J. Manag. Prod. 2019, 10, 1193–1212. [Google Scholar] [CrossRef]
- van Gasteren, H.; Krijgsveld, K.L.; Klauke, N.; Leshem, Y.; Metz, I.C.; Skakuj, M.; Sorbi, S.; Schekler, I.; Shamoun-Baranes, J. Aeroecology meets aviation safety: Early warning systems in Europe and the Middle East prevent collisions between birds and aircraft. Ecography 2019, 42, 899–911. [Google Scholar] [CrossRef]
- Bhusal, S.; Bhattarai, U.; Karkee, M. Improving Pest Bird Detection in a Vineyard Environment using Super-Resolution and Deep Learning. IFAC PapersOnLine 2019, 52, 18–23. [Google Scholar] [CrossRef]
- Gradolewski, D.; Dziak, D.; Kaniecki, D.; Jaworski, A.; Skakuj, M.; Kulesza, W.J. A Runway Safety System Based on Vertically Oriented Stereovision. Sensors 2021, 21, 1464. [Google Scholar] [CrossRef] [PubMed]
- Ballester, C.; Dupont, S.M.; Corbeau, A.; Chambert, T.; Duriez, O.; Besnard, A. A standardized protocol for assessing the performance of automatic detection systems used in onshore wind power plants to reduce avian mortality. J. Environ. Manag. 2024, 354, 120437. [Google Scholar] [CrossRef]
- Bruns, E.; Holger Ohlenburg, M.P. Kompetenzzentrum Naturschutz und Energiewende—Veröffentlichungen zu Antikollisionssystemen. Kompetenzzentrum Naturschutz Energiewende 2024. Available online: https://www.naturschutz-energiewende.de/fachwissen/veroeffentlichungen-zu-antikollisionssystemen (accessed on 28 February 2024).
- Gradolewski, D.; Dziak, D.; Martynow, M.; Kaniecki, D.; Szurlej-Kielanska, A.; Jaworski, A.; Kulesza, W.J. Comprehensive Bird Preservation at Wind Farms. Sensors 2021, 21, 267. [Google Scholar] [CrossRef]
- Duerr, A.E.; Parsons, A.E.; Nagy, L.R.; Kuehn, M.J.; Bloom, P.H. Effectiveness of an artificial intelligence-based system to curtail wind turbines to reduce eagle collisions. PLoS ONE 2023, 18, e0278754. [Google Scholar] [CrossRef]
- Bruns, E.; Ohlenburg, H.; Pommeranz, M. Kompetenzzentrum Naturschutz und Energiewende - Einsatz von Antikollisionssystemen unter Berücksichtigung der Wirtschaftlichen Zumutbarkeit. Kompetenzzentrum Naturschutz Energiewende 2024. Available online: https://www.naturschutz-energiewende.de/fachwissen/veroeffentlichungen/einsatz-von-antikollisionssystemen-unter-beruecksichtigung-der-wirtschaftlichen-zumutbarkeit (accessed on 28 February 2024).
- Rolek, B.W.; Braham, M.A.; Miller, T.A.; Duerr, A.E.; Katzner, T.E.; McClure, C.J.W. Variation in flight characteristics associated with entry by eagles into rotor-swept zones of wind turbines. IBIS 2024, 166, 308–314. [Google Scholar] [CrossRef]
- Elke Bruns, E.S.; Streiffeler, J. Anforderungen an technische Überwachungs- und Abschaltsysteme an Windenergieanlagen: Abschlussbericht der Workshopreihe “Technische Systeme”. Kompetenzzentrum Naturschutz Energiewende 2021. [Google Scholar] [CrossRef]
- Xie, J.; Zhong, Y.; Zhang, J.; Liu, S.; Ding, C.; Triantafyllopoulos, A. A review of automatic recognition technology for bird vocalizations in the deep learning era. Ecol. Inform. 2023, 73, 101927. [Google Scholar] [CrossRef]
- Principato, M.; Hasselwander, L.; Stangner, M.; Buettner, R. Unlocking the Potential of Wind Energy with Machine Learning-Based Avian Detection: A Call to Action. IEEE Access 2023, 11, 64026–64048. [Google Scholar] [CrossRef]
- Chan, J.J.X.; Srigrarom, S.; Cao, J.; Wang, P.; Ratsamee, P. Small Flying Object Classifications Based on Trajectories and Support Vector Machines. J. Robot. Mechatron. 2021, 33, 329–338. [Google Scholar] [CrossRef]
- Rahman, S.; Robertson, D.A. Classification of drones and birds using convolutional neural networks applied to radar micro-Doppler spectrogram images. IET Radar Sonar Navig. 2020, 14, 653–661. [Google Scholar] [CrossRef]
- Hernández-Pliego, J.; Rodríguez, C.; Dell’Omo, G.; Bustamante, J. Combined use of tri-axial accelerometers and GPS reveals the flexible foraging strategy of a bird in relation to weather conditions. PLoS ONE 2017, 12, e0177892. [Google Scholar] [CrossRef] [PubMed]
- Atanbori, J.; Duan, W.; Shaw, E.; Appiah, K.; Dickinson, P. Classification of bird species from video using appearance and motion features. Ecol. Inform. 2018, 48, 12–23. [Google Scholar] [CrossRef]
- Qian, Y. Exploration of machine algorithms based on deep learning model and feature extraction. Math. Biosci. Eng. 2021, 18, mbe–18–06–376. [Google Scholar] [CrossRef]
- Putrada, A.G.; Prabowo, S. ImbGAFS: GA Feature Selection for AUC in Bird Strike Prediction. In CS & IT Conference Proceedings of the 4th International Conference on Machine Learning Techniques and NLP (MLNLP 2023), Copenhagen, Denmark, 16–17 September 2023; AIRCC Publishing: Chennai, India, 2023; Volume 13, p. 13. [Google Scholar]
- Kumar, Y.; Gupta, S.; Singh, W. A novel deep transfer learning models for recognition of birds sounds in different environment. Soft Comput. 2022, 26, 1003–1023. [Google Scholar] [CrossRef]
- Kahl, S.; Navine, A.; Denton, T.; Klinck, H.; Hart, P.; Glotin, H.; Goëau, H.; Vellinga, W.P.; Planqué, R.; Joly, A. Overview of BirdCLEF 2022: Endangered bird species recognition in soundscape recordings. In Proceedings of the CLEF 2022: Conference and Labs of the Evaluation Forum, Bologna, Italy, 5–8 September 2022. Working Notes of CLEF. [Google Scholar]
- Zhang, X.; Mehta, V.; Bolic, M.; Mantegh, I. Hybrid AI-enabled Method for UAS and Bird Detection and Classification. In Proceedings of the 2020 IEEE International Conference on Systems, Man, and Cybernetics (SMC), Toronto, ON, Canada, 11–14 October 2020; pp. 2803–2807. [Google Scholar] [CrossRef]
- Qiao, B.; Zhou, Z.; Yang, H.; Cao, J. Bird species recognition based on SVM classifier and decision tree. In Proceedings of the 2017 1st International Conference on Electronics Instrumentation & Information Systems (EIIS), Harbin, China, 3–5 June 2017; pp. 1–4. [Google Scholar] [CrossRef]
- Xu, S.; Zhu, Q. Seabird image identification in natural scenes using Grabcut and combined features. Ecol. Inform. 2016, 33, 24–31. [Google Scholar] [CrossRef]
- Xie, J.; Zhu, M. Handcrafted features and late fusion with deep learning for bird sound classification. Ecol. Inform. 2019, 52, 74–81. [Google Scholar] [CrossRef]
- Atanbori, J.; Duan, W.; Murray, J.; Appiah, K.; Dickinson, P. Automatic classification of flying bird species using computer vision techniques. Pattern Recogn. Lett. 2016, 81, 53–62. [Google Scholar] [CrossRef]
- Roslan, R.; Nazery, N.A.; Jamil, N.; Hamzah, R. Color-based bird image classification using Support Vector Machine. In Proceedings of the 2017 IEEE 6th Global Conference on Consumer Electronics (GCCE), Nagoya, Japan, 24–27 October 2017; pp. 1–5. [Google Scholar] [CrossRef]
- Gradolewski, D.; Maslowski, D.; Dziak, D.; Jachimczyk, B.; Mundlamuri, S.T.; Prakash, C.G.; Kulesza, W.J. A Distributed Computing Real-Time Safety System of Collaborative Robot. Elektron. Elektrotech. 2020, 26, 4–14. [Google Scholar] [CrossRef]
- Luedicke, J. Friedman’s Super Smoother; Technical report; StataCorp: College Station, TX, USA, 2015. [Google Scholar]
- Pedregosa, F.; Varoquaux, G.; Gramfort, A.; Michel, V.; Thirion, B.; Grisel, O.; Blondel, M.; Prettenhofer, P.; Weiss, R.; Dubourg, V.; et al. Scikit-learn: Machine Learning in Python. J. Mach. Learn. Res. 2011, 12, 2825–2830. [Google Scholar]
- Gad, A.F. PyGAD: An Intuitive Genetic Algorithm Python Library. CoRR 2021, arXiv:2106.06158. [Google Scholar] [CrossRef]
Quantity | Symbol | Physical Unit | Features | ||
---|---|---|---|---|---|
Histogram with n-th Distance Interval | Average Value within n-th Distance Intervals | Variance | |||
Angular velocity | |||||
Polar angle | |||||
Size | |||||
Arc path length | |||||
Airplanes vs. Birds (RF, th = 0.4) | Big_Raptors vs. Other_Birds (RF, th = 0.125) | |||||
---|---|---|---|---|---|---|
Full Features | CCF | GA | Full Features | CCF | GA | |
(78 Features) | (60 Features) | (39 Features) | (78 Features) | (66 Features) | (38 Features) | |
Accuracy | 95.5% | 95.7% | 97% | 73.7% | 74.6% | 77.2% |
Recall | 98.6 % | 98.6% | 98.6% | 89.8% | 90.4% | 93.5% |
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content. |
© 2024 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https://creativecommons.org/licenses/by/4.0/).
Share and Cite
Tkaczyk, R.; Madejski, G.; Gradolewski, D.; Dziak, D.; Kulesza, W.J. Methodological Selection of Optimal Features for Object Classification Based on Stereovision System. Sensors 2024, 24, 3941. https://doi.org/10.3390/s24123941
Tkaczyk R, Madejski G, Gradolewski D, Dziak D, Kulesza WJ. Methodological Selection of Optimal Features for Object Classification Based on Stereovision System. Sensors. 2024; 24(12):3941. https://doi.org/10.3390/s24123941
Chicago/Turabian StyleTkaczyk, Rafał, Grzegorz Madejski, Dawid Gradolewski, Damian Dziak, and Wlodek J. Kulesza. 2024. "Methodological Selection of Optimal Features for Object Classification Based on Stereovision System" Sensors 24, no. 12: 3941. https://doi.org/10.3390/s24123941
APA StyleTkaczyk, R., Madejski, G., Gradolewski, D., Dziak, D., & Kulesza, W. J. (2024). Methodological Selection of Optimal Features for Object Classification Based on Stereovision System. Sensors, 24(12), 3941. https://doi.org/10.3390/s24123941