Anomalous_sound_detection_dacon

Dacon competitions

• 월간 데이콘 기계 고장 진단 AI 경진대회

How to Use

• Run main.py

Requirements

• pandas == 1.5.3
• numpy == 1.21.6
• torch == 1.13.1+cu116
• librosa == 0.9.2
• tqdm == 4.64.1

Metric

• Macro F1 Score

Score

• Public score : 86.07 (59/600)
• Private score : 84.40 (64/600)

Workers

노영준

• Seed ensemble
• Exploratory Data Analysis
• Code refactoring
• Project Managing
• AutoEncoder

이가은

• log-mel spectrogram
• find threshold method

Citation

• Two-stage anomalous sound detection systems using domain generalization and specialization techniques