Diana Ikasari
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Diana Ikasari has a total of 5 co-authors. Here are the top co-authors by number of shared publications:
- Achmad Benny Mutiara
Gunadarma University
1 shared publication - Dewi Agushinta Rahayu
Universitas Gunadarma
1 shared publication - Radityo Hendratmojo Jati Notonegoro
Universitas Gunadarma
1 shared publication - Rifki Kosasih
Universitas Gunadarma
1 shared publication - Witta Listiya Ningrum
Gunadarma University
1 shared publication
Latest Publications
- Open Access
- 1 October 2026
- Published by Institute of Advanced Engineering and Science in IAES International Journal of Artificial Intelligence (IJ-AI)
- Vol. 15 (5) , 4424-4431
Micro dataset segmentation for vanilla disease detection
Vanilla cultivation in Indonesia is often affected by various plant diseases, which pose significant challenges to productivity. However, images of vanilla disease data available on the internet are very limited for public access; the easily accessible disease images are mostly found in books related to vanilla or vanilla gardening. In this study, data collection involved 18 images of vanilla plants infected with basal stem rot caused by the fungus Sclerotium rolfsii and 20 images of healthy green vanilla plants sourced from websites. A pretrained U-Net model from “ImageNet” with a mobile network version 2 (MobileNetV2) backbone was employed, using parameters of 100 epochs, batch size 16, image size 256 × 256, and learning rate 0.0001. The model was trained using four different color preprocessing methods: RGB, CIE Lab, HSV, and YCbCr. Experimental results showed that the YCbCr color representations consistently achieved the best segmentation performance on both training and validation datasets. Color representations such as CIE Lab and YCbCr can better capture contrast and lighting variations compared to RGB, resulting in a more robust and generalizable model.
- 1 October 2026
- Published by Institute of Advanced Engineering and Science in IAES International Journal of Artificial Intelligence (IJ-AI)
- Vol. 15 (5) , 4424-4431
Open AccessMicro dataset segmentation for vanilla disease detection
Vanilla cultivation in Indonesia is often affected by various plant diseases, which pose significant challenges to productivity. However, images of vanilla disease data available on the internet are very limited for public access; the easily accessible disease images are mostly found in books related to vanilla or vanilla gardening. In this study, data collection involved 18 images of vanilla plants infected with basal stem rot caused by the fungus Sclerotium rolfsii and 20 images of healthy green vanilla plants sourced from websites. A pretrained U-Net model from “ImageNet” with a mobile network version 2 (MobileNetV2) backbone was employed, using parameters of 100 epochs, batch size 16, image size 256 × 256, and learning rate 0.0001. The model was trained using four different color preprocessing methods: RGB, CIE Lab, HSV, and YCbCr. Experimental results showed that the YCbCr color representations consistently achieved the best segmentation performance on both training and validation datasets. Color representations such as CIE Lab and YCbCr can better capture contrast and lighting variations compared to RGB, resulting in a more robust and generalizable model.
- Open Access
- 1 August 2026
- Published by Institute of Advanced Engineering and Science in IAES International Journal of Artificial Intelligence (IJ-AI)
- Vol. 15 (4) , 3365-3375
WVisionBERT-VL: a multimodal model architecture for toxicity classification on social media platforms using large language models
The increasing prevalence of toxic content on social media, conveyed through text, images, and videos, poses significant challenges for automated content moderation systems. Although prior studies have reported promising results in unimodal and bimodal settings, they often fail to capture implicit and contextual toxicity emerging from interactions across multiple modalities, particularly in non-English environments. This paper proposes WVisionBERT-VL, an end-to-end multimodal framework for toxicity detection that integrates text, image, and video modalities within a unified architecture. The proposed model incorporates modality-specific encoders, bidirectional multi-head cross attention (BMHCA) for cross-modal synchronization, and an adaptive fusion gate to dynamically balance modality contributions. A balanced multimodal dataset is constructed from social media platforms, including X, Instagram, and TikTok, and refined using a model-based labeling strategy with limited human-in-the-loop validation. Experimental results on a custom Indonesian dataset demonstrate strong in-domain performance, achieving an accuracy of 94.12%, a macro-F1 of 0.9407, and a receiver operating characteristic - area under the curve (ROC-AUC) of 0.9721, with robustness further validated through five-fold cross-validation. Cross-dataset evaluation highlights challenges related to domain shift, underscoring the need for future research on robust and domain-adaptive multimodal toxicity detection.
- 1 August 2026
- Published by Institute of Advanced Engineering and Science in IAES International Journal of Artificial Intelligence (IJ-AI)
- Vol. 15 (4) , 3365-3375
Open AccessWVisionBERT-VL: a multimodal model architecture for toxicity classification on social media platforms using large language models
The increasing prevalence of toxic content on social media, conveyed through text, images, and videos, poses significant challenges for automated content moderation systems. Although prior studies have reported promising results in unimodal and bimodal settings, they often fail to capture implicit and contextual toxicity emerging from interactions across multiple modalities, particularly in non-English environments. This paper proposes WVisionBERT-VL, an end-to-end multimodal framework for toxicity detection that integrates text, image, and video modalities within a unified architecture. The proposed model incorporates modality-specific encoders, bidirectional multi-head cross attention (BMHCA) for cross-modal synchronization, and an adaptive fusion gate to dynamically balance modality contributions. A balanced multimodal dataset is constructed from social media platforms, including X, Instagram, and TikTok, and refined using a model-based labeling strategy with limited human-in-the-loop validation. Experimental results on a custom Indonesian dataset demonstrate strong in-domain performance, achieving an accuracy of 94.12%, a macro-F1 of 0.9407, and a receiver operating characteristic - area under the curve (ROC-AUC) of 0.9721, with robustness further validated through five-fold cross-validation. Cross-dataset evaluation highlights challenges related to domain shift, underscoring the need for future research on robust and domain-adaptive multimodal toxicity detection.