maximegiguere/Dopamine-Analyzer

import streamlit as st import numpy as np import librosa import librosa.display import cv2 import tempfile import os from deepface import DeepFace

📌 Fonction pour calculer le score de dopamine

def dopamine_score(video_metrics): """ Calcule le score de dopamine d'une vidéo en fonction de plusieurs critères. """ weights = { "hook_strength": 20, "cuts_per_sec": 15, "emotion_intensity": 15, "cliffhangers": 10, "dopamine_words": 10, "visual_stimulation": 10, "cta_strength": 10, "narrative_structure": 5, "audio_intensity": 10, "facial_expressions": 10, "retention_rate": 10 } normalized_metrics = {key: min(1, max(0, video_metrics.get(key, 0))) for key in weights.keys()} total_score = sum(normalized_metrics[key] * weights[key] for key in weights) max_possible_score = sum(weights.values()) dopamine_score = round((total_score / max_possible_score) * 100, 2) return dopamine_score, normalized_metrics

📌 Analyse de l'audio pour évaluer l'intensité

def analyze_audio(audio_path): try: y, sr = librosa.load(audio_path) intensity = np.mean(librosa.feature.rms(y=y)) return min(1, intensity * 10) except Exception: return 0 # Retourne un score faible si l'analyse échoue

📌 Analyse des expressions faciales dans la vidéo

def analyze_facial_expressions(video_path): cap = cv2.VideoCapture(video_path) emotions_detected = [] while cap.isOpened(): ret, frame = cap.read() if not ret: break try: analysis = DeepFace.analyze(frame, actions=['emotion'], enforce_detection=False) emotions_detected.append(analysis[0]['dominant_emotion']) except: pass cap.release() emotion_strength = emotions_detected.count('happy') + emotions_detected.count('surprise') return min(1, emotion_strength / len(emotions_detected) if emotions_detected else 0)

📌 Analyse complète de la vidéo

def analyze_video_metrics(video_path, audio_path): video_data = { "hook_strength": 0.9, "cuts_per_sec": 0.8, "emotion_intensity": 0.7, "cliffhangers": 0.6, "dopamine_words": 0.7, "visual_stimulation": 0.8, "cta_strength": 0.9, "narrative_structure": 0.7, "audio_intensity": analyze_audio(audio_path), "facial_expressions": analyze_facial_expressions(video_path), "retention_rate": 0.8 } return dopamine_score(video_data)

📌 Interface Streamlit

st.title("📊 Dopamine Score Analyzer for Videos")

📌 Upload des fichiers vidéo et audio

uploaded_video = st.file_uploader("Upload your video file", type=["mp4", "mov", "avi"]) uploaded_audio = st.file_uploader("Upload the corresponding audio file", type=["wav", "mp3"])

if uploaded_video and uploaded_audio: with tempfile.NamedTemporaryFile(delete=False, suffix=".mp4") as temp_video: temp_video.write(uploaded_video.read()) video_path = temp_video.name with tempfile.NamedTemporaryFile(delete=False, suffix=".wav") as temp_audio: temp_audio.write(uploaded_audio.read()) audio_path = temp_audio.name

score, details = analyze_video_metrics(video_path, audio_path)

st.subheader(f"🔥 Dopamine Score: {score}/100")
for key, value in details.items():
    st.write(f"{key.replace('_', ' ').title()}: {value * 100}%")

os.unlink(video_path)
os.unlink(audio_path)