A Comparative Study of Preprocessing and Model Compression Techniques in Deep Learning for Forest Sound Classification
Deep-learning models play a significant role in modern software solutions, with the capabilities of handling complex tasks, improving accuracy, automating processes, and adapting to diverse domains, eventually contributing to advancements in various industries. This study provides a comparative stud...
| Main Authors: | , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2024-02-01
|
| Series: | Sensors |
| Subjects: | |
| Online Access: | https://www.mdpi.com/1424-8220/24/4/1149 |
| _version_ | 1797296994765504512 |
|---|---|
| author | Thivindu Paranayapa Piumini Ranasinghe Dakshina Ranmal Dulani Meedeniya Charith Perera |
| author_facet | Thivindu Paranayapa Piumini Ranasinghe Dakshina Ranmal Dulani Meedeniya Charith Perera |
| author_sort | Thivindu Paranayapa |
| collection | DOAJ |
| description | Deep-learning models play a significant role in modern software solutions, with the capabilities of handling complex tasks, improving accuracy, automating processes, and adapting to diverse domains, eventually contributing to advancements in various industries. This study provides a comparative study on deep-learning techniques that can also be deployed on resource-constrained edge devices. As a novel contribution, we analyze the performance of seven Convolutional Neural Network models in the context of data augmentation, feature extraction, and model compression using acoustic data. The results show that the best performers can achieve an optimal trade-off between model accuracy and size when compressed with weight and filter pruning followed by 8-bit quantization. In adherence to the study workflow utilizing the forest sound dataset, MobileNet-v3-small and ACDNet achieved accuracies of 87.95% and 85.64%, respectively, while maintaining compact sizes of 243 KB and 484 KB, respectively. Henceforth, this study concludes that CNNs can be optimized and compressed to be deployed in resource-constrained edge devices for classifying forest environment sounds. |
| first_indexed | 2024-03-07T22:14:47Z |
| format | Article |
| id | doaj.art-3f1bcd89f26043e2b237d5a832666a80 |
| institution | Directory Open Access Journal |
| issn | 1424-8220 |
| language | English |
| last_indexed | 2024-03-07T22:14:47Z |
| publishDate | 2024-02-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Sensors |
| spelling | doaj.art-3f1bcd89f26043e2b237d5a832666a802024-02-23T15:33:42ZengMDPI AGSensors1424-82202024-02-01244114910.3390/s24041149A Comparative Study of Preprocessing and Model Compression Techniques in Deep Learning for Forest Sound ClassificationThivindu Paranayapa0Piumini Ranasinghe1Dakshina Ranmal2Dulani Meedeniya3Charith Perera4Department of Computer Science & Engineering, University of Moratuwa, Moratuwa 10400, Sri LankaDepartment of Computer Science & Engineering, University of Moratuwa, Moratuwa 10400, Sri LankaDepartment of Computer Science & Engineering, University of Moratuwa, Moratuwa 10400, Sri LankaDepartment of Computer Science & Engineering, University of Moratuwa, Moratuwa 10400, Sri LankaSchool of Computer Science and Informatics, Cardiff University, Cardiff CF24 3AA, UKDeep-learning models play a significant role in modern software solutions, with the capabilities of handling complex tasks, improving accuracy, automating processes, and adapting to diverse domains, eventually contributing to advancements in various industries. This study provides a comparative study on deep-learning techniques that can also be deployed on resource-constrained edge devices. As a novel contribution, we analyze the performance of seven Convolutional Neural Network models in the context of data augmentation, feature extraction, and model compression using acoustic data. The results show that the best performers can achieve an optimal trade-off between model accuracy and size when compressed with weight and filter pruning followed by 8-bit quantization. In adherence to the study workflow utilizing the forest sound dataset, MobileNet-v3-small and ACDNet achieved accuracies of 87.95% and 85.64%, respectively, while maintaining compact sizes of 243 KB and 484 KB, respectively. Henceforth, this study concludes that CNNs can be optimized and compressed to be deployed in resource-constrained edge devices for classifying forest environment sounds.https://www.mdpi.com/1424-8220/24/4/1149augmentationfeature extractionclassificationpruningquantization |
| spellingShingle | Thivindu Paranayapa Piumini Ranasinghe Dakshina Ranmal Dulani Meedeniya Charith Perera A Comparative Study of Preprocessing and Model Compression Techniques in Deep Learning for Forest Sound Classification Sensors augmentation feature extraction classification pruning quantization |
| title | A Comparative Study of Preprocessing and Model Compression Techniques in Deep Learning for Forest Sound Classification |
| title_full | A Comparative Study of Preprocessing and Model Compression Techniques in Deep Learning for Forest Sound Classification |
| title_fullStr | A Comparative Study of Preprocessing and Model Compression Techniques in Deep Learning for Forest Sound Classification |
| title_full_unstemmed | A Comparative Study of Preprocessing and Model Compression Techniques in Deep Learning for Forest Sound Classification |
| title_short | A Comparative Study of Preprocessing and Model Compression Techniques in Deep Learning for Forest Sound Classification |
| title_sort | comparative study of preprocessing and model compression techniques in deep learning for forest sound classification |
| topic | augmentation feature extraction classification pruning quantization |
| url | https://www.mdpi.com/1424-8220/24/4/1149 |
| work_keys_str_mv | AT thivinduparanayapa acomparativestudyofpreprocessingandmodelcompressiontechniquesindeeplearningforforestsoundclassification AT piuminiranasinghe acomparativestudyofpreprocessingandmodelcompressiontechniquesindeeplearningforforestsoundclassification AT dakshinaranmal acomparativestudyofpreprocessingandmodelcompressiontechniquesindeeplearningforforestsoundclassification AT dulanimeedeniya acomparativestudyofpreprocessingandmodelcompressiontechniquesindeeplearningforforestsoundclassification AT charithperera acomparativestudyofpreprocessingandmodelcompressiontechniquesindeeplearningforforestsoundclassification AT thivinduparanayapa comparativestudyofpreprocessingandmodelcompressiontechniquesindeeplearningforforestsoundclassification AT piuminiranasinghe comparativestudyofpreprocessingandmodelcompressiontechniquesindeeplearningforforestsoundclassification AT dakshinaranmal comparativestudyofpreprocessingandmodelcompressiontechniquesindeeplearningforforestsoundclassification AT dulanimeedeniya comparativestudyofpreprocessingandmodelcompressiontechniquesindeeplearningforforestsoundclassification AT charithperera comparativestudyofpreprocessingandmodelcompressiontechniquesindeeplearningforforestsoundclassification |