In-The-Wild deepfake detection using adaptable CNN models with visual class activation mapping for improved accuracy
Deepfake technology has become increasingly sophisticated in recent years, making detecting fake images and videos challenging. This paper investigates the performance of adaptable convolutional neural network (CNN) models for detecting Deepfakes. In-the-wild OpenForensics dataset was used to evalua...
| Main Authors: | , , , |
|---|---|
| Format: | Conference or Workshop Item |
| Language: | English English |
| Published: |
2023
|
| Subjects: | |
| Online Access: | http://umpir.ump.edu.my/id/eprint/38714/1/In-the-wild%20deepfake%20detection%20using%20adaptable%20CNN%20models.pdf http://umpir.ump.edu.my/id/eprint/38714/2/In-The-Wild%20deepfake%20detection%20using%20adaptable%20CNN%20models%20with%20visual%20class%20activation%20mapping%20for%20improved%20accuracy_ABS.pdf |
| _version_ | 1796995916310249472 |
|---|---|
| author | Muhammad Salihin, Saealal Mohd Zamri, Ibrahim Mohd Ibrahim, Shapiai Norasyikin, Fadilah |
| author_facet | Muhammad Salihin, Saealal Mohd Zamri, Ibrahim Mohd Ibrahim, Shapiai Norasyikin, Fadilah |
| author_sort | Muhammad Salihin, Saealal |
| collection | UMP |
| description | Deepfake technology has become increasingly sophisticated in recent years, making detecting fake images and videos challenging. This paper investigates the performance of adaptable convolutional neural network (CNN) models for detecting Deepfakes. In-the-wild OpenForensics dataset was used to evaluate four different CNN models (DenseNet121, ResNet18, SqueezeNet, and VGG11) at different batch sizes and with various performance metrics. Results show that the adapted VGG11 model with a batch size of 32 achieved the highest accuracy of 94.46% in detecting Deepfakes, outperforming the other models, with DenseNet121 as the second-best performer achieving an accuracy of 93.89% with the same batch size. Grad-CAM techniques are utilized to visualize the decision-making process within the models, aiding in understanding the Deepfake classification process. These findings provide valuable insights into the performance of different deep learning models and can guide the selection of an appropriate model for a specific application. |
| first_indexed | 2024-03-06T13:09:19Z |
| format | Conference or Workshop Item |
| id | UMPir38714 |
| institution | Universiti Malaysia Pahang |
| language | English English |
| last_indexed | 2024-03-06T13:09:19Z |
| publishDate | 2023 |
| record_format | dspace |
| spelling | UMPir387142023-11-06T04:24:15Z http://umpir.ump.edu.my/id/eprint/38714/ In-The-Wild deepfake detection using adaptable CNN models with visual class activation mapping for improved accuracy Muhammad Salihin, Saealal Mohd Zamri, Ibrahim Mohd Ibrahim, Shapiai Norasyikin, Fadilah T Technology (General) TA Engineering (General). Civil engineering (General) TK Electrical engineering. Electronics Nuclear engineering Deepfake technology has become increasingly sophisticated in recent years, making detecting fake images and videos challenging. This paper investigates the performance of adaptable convolutional neural network (CNN) models for detecting Deepfakes. In-the-wild OpenForensics dataset was used to evaluate four different CNN models (DenseNet121, ResNet18, SqueezeNet, and VGG11) at different batch sizes and with various performance metrics. Results show that the adapted VGG11 model with a batch size of 32 achieved the highest accuracy of 94.46% in detecting Deepfakes, outperforming the other models, with DenseNet121 as the second-best performer achieving an accuracy of 93.89% with the same batch size. Grad-CAM techniques are utilized to visualize the decision-making process within the models, aiding in understanding the Deepfake classification process. These findings provide valuable insights into the performance of different deep learning models and can guide the selection of an appropriate model for a specific application. 2023 Conference or Workshop Item PeerReviewed pdf en http://umpir.ump.edu.my/id/eprint/38714/1/In-the-wild%20deepfake%20detection%20using%20adaptable%20CNN%20models.pdf pdf en http://umpir.ump.edu.my/id/eprint/38714/2/In-The-Wild%20deepfake%20detection%20using%20adaptable%20CNN%20models%20with%20visual%20class%20activation%20mapping%20for%20improved%20accuracy_ABS.pdf Muhammad Salihin, Saealal and Mohd Zamri, Ibrahim and Mohd Ibrahim, Shapiai and Norasyikin, Fadilah (2023) In-The-Wild deepfake detection using adaptable CNN models with visual class activation mapping for improved accuracy. In: 2023 5th International Conference on Computer Communication and the Internet, ICCCI 2023 , 23-25 June 2023 , Fujisawa. pp. 9-14.. ISBN 979-835032695-6 https://doi.org/10.1109/ICCCI59363.2023.10210096 |
| spellingShingle | T Technology (General) TA Engineering (General). Civil engineering (General) TK Electrical engineering. Electronics Nuclear engineering Muhammad Salihin, Saealal Mohd Zamri, Ibrahim Mohd Ibrahim, Shapiai Norasyikin, Fadilah In-The-Wild deepfake detection using adaptable CNN models with visual class activation mapping for improved accuracy |
| title | In-The-Wild deepfake detection using adaptable CNN models with visual class activation mapping for improved accuracy |
| title_full | In-The-Wild deepfake detection using adaptable CNN models with visual class activation mapping for improved accuracy |
| title_fullStr | In-The-Wild deepfake detection using adaptable CNN models with visual class activation mapping for improved accuracy |
| title_full_unstemmed | In-The-Wild deepfake detection using adaptable CNN models with visual class activation mapping for improved accuracy |
| title_short | In-The-Wild deepfake detection using adaptable CNN models with visual class activation mapping for improved accuracy |
| title_sort | in the wild deepfake detection using adaptable cnn models with visual class activation mapping for improved accuracy |
| topic | T Technology (General) TA Engineering (General). Civil engineering (General) TK Electrical engineering. Electronics Nuclear engineering |
| url | http://umpir.ump.edu.my/id/eprint/38714/1/In-the-wild%20deepfake%20detection%20using%20adaptable%20CNN%20models.pdf http://umpir.ump.edu.my/id/eprint/38714/2/In-The-Wild%20deepfake%20detection%20using%20adaptable%20CNN%20models%20with%20visual%20class%20activation%20mapping%20for%20improved%20accuracy_ABS.pdf |
| work_keys_str_mv | AT muhammadsalihinsaealal inthewilddeepfakedetectionusingadaptablecnnmodelswithvisualclassactivationmappingforimprovedaccuracy AT mohdzamriibrahim inthewilddeepfakedetectionusingadaptablecnnmodelswithvisualclassactivationmappingforimprovedaccuracy AT mohdibrahimshapiai inthewilddeepfakedetectionusingadaptablecnnmodelswithvisualclassactivationmappingforimprovedaccuracy AT norasyikinfadilah inthewilddeepfakedetectionusingadaptablecnnmodelswithvisualclassactivationmappingforimprovedaccuracy |