From 3b4c1d632c0f7b252be500a863b1203e6f1dac2c Mon Sep 17 00:00:00 2001
From: Nishant Dwivedi <135944619+nishant4500@users.noreply.github.com>
Date: Sat, 19 Oct 2024 22:36:38 +0530
Subject: [PATCH] Create recognized_gesture.py

---
 Detection Models/recognized_gesture.py | 82 ++++++++++++++++++++++++++
 1 file changed, 82 insertions(+)
 create mode 100644 Detection Models/recognized_gesture.py

diff --git a/Detection Models/recognized_gesture.py b/Detection Models/recognized_gesture.py
new file mode 100644
index 000000000..8309a723e
--- /dev/null
+++ b/Detection Models/recognized_gesture.py	
@@ -0,0 +1,82 @@
+import cv2
+from flask import Flask, render_template
+import cv2
+import openai
+import mediapipe as mp
+import pyttsx3
+from langchain import ChatOpenAI
+
+# Initialize OpenAI API
+openai.api_key = "YOUR_OPENAI_API_KEY"
+
+# Initialize text-to-speech
+engine = pyttsx3.init()
+
+# Function to send gesture to LLM for context
+def interpret_gesture(gesture_description):
+    prompt = f"What does the gesture '{gesture_description}' signify?"
+    response = openai.Completion.create(
+        engine="text-davinci-003",
+        prompt=prompt,
+        max_tokens=60
+    )
+    return response.choices[0].text.strip()
+
+# Function for voice output
+def speak_text(text):
+    engine.say(text)
+    engine.runAndWait()
+
+# Initialize MediaPipe Hands for gesture tracking
+mp_hands = mp.solutions.hands
+hands = mp_hands.Hands()
+mp_draw = mp.solutions.drawing_utils
+
+# Webcam capture
+cap = cv2.VideoCapture(0)
+
+while cap.isOpened():
+    ret, frame = cap.read()
+    if not ret:
+        break
+
+    # Convert the frame to RGB for MediaPipe
+    rgb_frame = cv2.cvtColor(frame, cv2.COLOR_BGR2RGB)
+    result = hands.process(rgb_frame)
+
+    if result.multi_hand_landmarks:
+        for hand_landmarks in result.multi_hand_landmarks:
+            mp_draw.draw_landmarks(frame, hand_landmarks, mp_hands.HAND_CONNECTIONS)
+
+            # Here you would add logic to classify the hand gesture
+            # For simplicity, let's assume we recognize a "thumbs up"
+            recognized_gesture = "thumbs up"
+            gesture_meaning = interpret_gesture(recognized_gesture)
+
+            print(f"Recognized Gesture: {recognized_gesture}")
+            print(f"Interpreted Meaning: {gesture_meaning}")
+
+            # Voice output
+            speak_text(f"Gesture: {recognized_gesture}. Meaning: {gesture_meaning}")
+
+    # Display the frame
+    cv2.imshow("Gesture Prediction", frame)
+
+    # Exit with the 'q' key
+    if cv2.waitKey(10) & 0xFF == ord('q'):
+        break
+
+cap.release()
+cv2.destroyAllWindows()
+
+
+app = Flask(__name__)
+
+@app.route('/')
+def index():
+    return render_template('index.html')
+
+# Gesture recognition logic can be linked here to update the UI
+
+if __name__ == '__main__':
+    app.run(debug=True)