Algorithm 1 End-to-end workflow of AGREE-YOLO

Input: Raw seafood image I, user natural-language query Q Output: Culturally adapted recipe R, dish image D

1:   B ← YOLOv13_GSConv_WIoU (I)           → Bounding boxes & class labels 2:   for each detection b in B do

3:          confidence ← b.confidence 4:          if confidence ≥ 0.83 then

5:      store_into_MySQL (b.class_label, b.confidence, timestamp) 6:          end if 7:   end for 8:   refined_query ← USRR (Q)            → User Seafood Requirement Refinemen 9:   SQL_stmt ← rookie_text2data (refined_query)       → Convert NL to SQL 10: records ← rookie_execute_sql (SQL_stmt)     → Fetch from MySQ 11: cuisine ← CCS (records)              → Conditional Country Selection 12: R ← CMG (records, cuisine)            → Cooking Method Generato 13: D ← SDP (records, cuisine)           → Seafood Dish Picture Generator 14:   return (R, D)