Maldini, Andry Syva (2025) IMPLEMENTASI FASTER R-CNN DAN OCR DENGAN PENDEKATAN NATURAL LANGUAGE PROCESSING UNTUK DETEKSI IKLAN JUDI ONLINE. Undergraduate thesis, Universitas Pembangunan Nasional Veteran Jawa Timur.
![]() |
Text (COVER)
Cover.pdf Download (1MB) |
![]() |
Text (BAB I)
21083010085_BAB I.pdf Download (142kB) |
![]() |
Text (BAB II)
21083010085_BAB II.pdf Restricted to Repository staff only until 28 July 2028. Download (756kB) |
![]() |
Text (BAB III)
21083010085_BAB III.pdf Restricted to Repository staff only until 28 July 2028. Download (475kB) |
![]() |
Text (BAB IV)
21083010085_BAB IV.pdf Restricted to Repository staff only until 28 July 2028. Download (1MB) |
![]() |
Text (BAB V)
21083010085_BAB V.pdf Restricted to Repository staff only until 28 July 2028. Download (11kB) |
![]() |
Text (DAFTAR PUSTAKA)
21083010085_DAFTAR PUSTAKA.pdf Download (236kB) |
![]() |
Text (LAMPIRAN)
21083010085_LAMPIRAN.pdf Restricted to Repository staff only until 28 July 2028. Download (56kB) |
Abstract
The proliferation of online gambling advertisements on social media has increasingly adopted covert strategies such as visual watermarks and textual euphemisms, making manual content moderation ineffective. This research aims to develop an automated detection system for identifying online gambling advertisements using an end-to-end artificial intelligence approach. The methodology involves object detection using Faster R-CNN (ResNet-50 + FPN), text extraction using Transformer-based Optical Character Recognition (TrOCR), and semantic classification using BERT-based Natural Language Processing. The dataset was collected and annotated manually, followed by augmentation using Roboflow. Evaluation results show that the Faster R-CNN model achieved a mAP@0.50 of 98.1%, TrOCR recorded a Character Error Rate (CER) of 4.6% and a Word Error Rate (WER) of 29%, while the BERT classification model achieved 99% accuracy, with high precision and recall for both classes. These components were integrated into a Flask-based web pipeline, allowing users to upload images or short videos for automatic detection and classification. The system effectively detects both explicit and obfuscated gambling advertisements and provides a practical tool to support content moderation efforts on digital platforms.
Item Type: | Thesis (Undergraduate) | ||||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|
Contributors: |
|
||||||||||||
Subjects: | H Social Sciences > HM Sociology H Social Sciences > HN Social history and conditions. Social problems. Social reform Q Science > QA Mathematics > QA76.6 Computer Programming |
||||||||||||
Divisions: | Faculty of Computer Science > Departemen of Data Science | ||||||||||||
Depositing User: | Andry Syva Maldini | ||||||||||||
Date Deposited: | 05 Aug 2025 07:58 | ||||||||||||
Last Modified: | 05 Aug 2025 07:58 | ||||||||||||
URI: | https://repository.upnjatim.ac.id/id/eprint/41565 |
Actions (login required)
![]() |
View Item |