Mobile-Based Translation System for Cebuano Language with Object Detection for Travel Assistance using Neural Machine Translation

The Philippines as an archipelagic nation has the numerous number of dialects per island that hampers local tourist communication. Cebu is one of the top tourist destinations by locals also has a vast language hurdle compared to other local dialects. Because of huge language differences of native Cebuano and Filipino speakers, travelers in Cebu often experience difficulty in expressing their needs and requests to the island locals. The solution is a mobile-based translation system with object detection for travel assistance. The system uses Neural Machine Translation to translate Filipino to Cebuano language and vice versa based on the user's keyboard input, extracted text strings from a detected image, and the detected object itself. The Filipino-Cebuano model obtained 31.1 BLEU score, while the Cebuano-Filipino model obtained 31.6 BLEU score.