Merge pull request #15 from unicam-complex-system/object-detection

Machine Learning algorithm completed

ImageRecognition/README.md

@@ -0,0 +1,24 @@
+## Technology
+<p align="center">
+  <a href="https://www.ultralytics.com" target="blank"><img src="https://assets-global.website-files.com/646dd1f1a3703e451ba81ecc/64777c3e071ec953437e6950_logo.svg"  alt="Ultralytics Yolo" /></a>
+</p>
+
+Ultralytics YOLO for building accurate and efficient object detection models.
+
+## Running the app
+First, open the terminal and move to the ImageRecognition directiory.
+Run send_frame.py for sending camera frames:
+
+```bash
+python send_frame.py
+```
+
+To start the object detection function move to the src directory and run ObjectDetectionYOLO.py:
+
+```bash
+python ObjectDetectionYOLO.py
+```
+
+## Learn More
+To learn more about YOLO architecture, take a look at the following resource:
+- [YOLO Documentation](https://docs.ultralytics.com).

ImageRecognition/requirements.txt

@@ -1,4 +1,5 @@
 opencv-python~=4.8.1.78
 numpy~=1.26.1
 ping3~=4.0.4
-ffmpeg-python~=0.2.0
+python>=3.8
+PyTorch>=1.8

ImageRecognition/send_frame.py

@@ -1,6 +1,5 @@
 import cv2
 import socketio
-import base64
 import threading
 
 #cameras data
@@ -40,7 +39,7 @@ def send_frames(camera_id, capture, interval=0.1):
 for camera_id, camera_url in cameras.items():
     capture = cv2.VideoCapture(camera_url)
     if not capture.isOpened():
-        print(f"Errore nell'apertura della telecamera {camera_id}")
+        print(f"Error stream with camera {camera_id}")
         continue
 
     threading.Thread(target=send_frames, args=(camera_id, capture)).start()

ImageRecognition/src/ObjectDetectionYOLO.py

@@ -30,18 +30,21 @@
 cam8 = 'rtsp://192.168.1.41:80/ch7_0.264'
 
 
-TOKEN = '6929905186:AAEouI3G2sbfS-y6ZzkXrpNgPQRAPs5_v-g'
-channel_id = '792557360'
 last_sent_image = None
 
-def post_request(image):
+def post_request(image, camera_id, status):
     global last_sent_image
-    apiURL = f"https://api.telegram.org/bot{TOKEN}/sendPhoto?chat_id={channel_id}"
+    url = 'http://localhost:8080'
+
+    data = {
+        'id': camera_id,
+        'status': status,
+    }
 
     if last_sent_image is None:
         last_sent_image = image
         try:
-            response = requests.post(apiURL, files={'photo':  image})
+            response = requests.post(url, data=data, files={'file':  (image, 'image/jpeg')})
             print("Status code is: ", response.status_code)      
         except requests.exceptions.RequestException as e:
             print("There was an exception that occurred while handling your request.", e)
@@ -59,7 +62,7 @@ def post_request(image):
         print("MEAN SQUARED ERROR is: ", mse)
         if mse > 105.85:
             try:
-                response = requests.post(apiURL, files={'photo':  image})
+                response = requests.post(url, data=data, files={'file':  (image, 'image/jpeg')})
                 print("Status code is: ", response.status_code)
 
                 if response.status_code == 200:
@@ -73,16 +76,17 @@ def post_request(image):
 #filename = url of cam
 #file_index = index of the file that can be assigned to each thread. cam1 has file_index as 1, cam2 has file_index as 2...
 def detection(filename, file_index):
-    cap = cv2.VideoCapture(filename)  # Read the video file
+    cap = cv2.VideoCapture(filename) 
     global last_sent_image
+    status = 'offline'
 
     while cap.isOpened():
 
        success, img = cap.read()
-       # Exit the loop if no more frames in either video
        if not success:
            break
 
+       status = 'online'
        results = model(img, stream = True, classes = 0, conf=0.5) 
        foundPerson = False
        #coordinates
@@ -91,8 +95,7 @@ def detection(filename, file_index):
           for box in boxes:
             #bounding box
             x1, y1, x2, y2 = box.xyxy[0]
-            x1, y1, x2, y2 = int(x1), int(y1), int(x2), int(y2) # convert to int values
-            # put box in cam
+            x1, y1, x2, y2 = int(x1), int(y1), int(x2), int(y2)
             cv2.rectangle(img, (x1, y1), (x2, y2), (255, 0, 255), 3)
             # confidence
             confidence = math.ceil((box.conf[0]*100))/100
@@ -105,14 +108,13 @@ def detection(filename, file_index):
 
        if(foundPerson):
            _, img_encoded = cv2.imencode('.jpg', img)
-           post_request(img_encoded)
+           post_request(img_encoded, file_index, status)
 
 
     cap.release()
 
 
 
-#initialize thread
 detect_thread1 = threading.Thread(target=detection,
                                    args=(cam1, 1),
                                    daemon=True)
@@ -145,7 +147,7 @@ def detection(filename, file_index):
                                    args=(cam8, 8),
                                    daemon=True)
 
-#start thread
+
 detect_thread1.start()
 detect_thread2.start()
 detect_thread3.start()