(Web Browser Interface)
(Application Logic)
EasyOCR
(English Text Extraction)
Text Preprocessing
(Cleaning + Normalization)
Clean English Caption
(Final OCR Output)
mBART Tokenizer
(Tokenization + Lang Tag)
mBART-50 Model
(Multilingual Translation)
Tamil / Telugu / Malayalam
Translation
BART-MNLI Classifier
Scene / Context Detection
(e.g., “animal”)
RESULT PAGE
• Uploaded Image
• English Extracted Caption
• Translated Output
• Scene / Context Label
by Tanuja