added finished test track and botsort yaml

Test_Realtime_label.py

@@ -1,12 +1,35 @@
 import cv2
 from ultralytics import YOLO
 from collections import deque
+import paho.mqtt.client as mqtt
+from influxdb import InfluxDBClient
+from influxdb_client import InfluxDBClient, Point, WriteOptions
+import time
+from datetime import datetime
+
+# InfluxDB Configuration
+INFLUX_URL = "http://localhost:8086"
+INFLUX_TOKEN = "--k98NX5UQ2qBCGAO80lLc_-teD-AUtKNj4uQfz0M8WyjHt04AT9d0dr6w8pup93ukw6YcJxWURmo2v6CAP_2g=="
+INFLUX_ORG = "GAAIM"
+INFLUX_BUCKET = "AGVIGNETTE"
+
+# Connect to InfluxDB
+client = InfluxDBClient(url=INFLUX_URL, token=INFLUX_TOKEN, org=INFLUX_ORG)
+write_api = client.write_api(write_options=WriteOptions(batch_size=1))
+
+# MQTT Setup
+MQTT_BROKER = "192.168.10.51"
+MQTT_TOPIC = "fruit/classification"
+
+mqtt_client = mqtt.Client()
+mqtt_client.connect(MQTT_BROKER, 1883, 6000)
 
 # Camera index (default camera is 0)
-camera_index = 1
+camera_index = 0
+i = 0
 
 # Load the YOLO model
-model = YOLO(r"D:\AIM\lemon\runs\detect\train4\weights\best.pt")  # Load custom model
+model = YOLO(r"/Users/vel/Desktop/CvModel/CV_AG/runs/detect/train5/weights/best.pt")  # Load custom model
 
 # Initialize the camera
 cap = cv2.VideoCapture(camera_index)
@@ -32,8 +55,9 @@ class_labels = {
 id_tracked_classes = ["DefectiveLemon", "GoodLemon", "NotRipeLemon"]
 
 # Parameters
-HISTORY_LENGTH = 5  # Number of frames to consider for majority voting
+HISTORY_LENGTH = 7  # Number of frames to consider for majority voting
-CONFIRMATION_FRAMES = 5  # Frames needed to confirm a new label
+CONFIRMATION_FRAMES = 7  # Frames needed to confirm a new label
+lemon_time = 0
 
 # Dictionary to track detection history and confirmed states
 lemon_history = {}  # Format: {ID: deque(maxlen=HISTORY_LENGTH)}
@@ -76,7 +100,7 @@ while cap.isOpened():
         break
 
     # Perform object tracking using BoT-SORT
-    results = model.track(source=frame, conf=0.5, tracker='botsort.yaml', show=False)
+    results = model.track(source=frame, conf=0.5, tracker='botsort.yaml', show=False, device = 'mps')
 
     for result in results:
         frame = result.orig_img  # Current frame
@@ -109,6 +133,43 @@ while cap.isOpened():
             cv2.rectangle(frame, (x1, y1), (x2, y2), (0, 255, 0), 2)
             cv2.putText(frame, display_text, (x1, y1 - 10),
                         cv2.FONT_HERSHEY_SIMPLEX, 0.6, (0, 255, 0), 2)
+            
+            # Create Decision Point at x = 600
+            if x1 > 100:
+                cv2.line(frame, (600, 0), (600, height), (255, 0, 0), 2)
+            # Create Decision Point at x = 670
+            if x1 > 100:
+                cv2.line(frame, (760, 0), (760, height), (255, 0, 0), 2)
+                cv2.putText(frame, "Decision Point", (630, height // 2),
+                            cv2.FONT_HERSHEY_SIMPLEX, 1, (255, 0, 0), 2)
+            
+            # Lock in the label once it crosses the decision point
+            if x1 > 700 and obj_id in lemon_states:
+                cv2.putText(frame, f"Locked: {lemon_states[obj_id]}", (x1, y1 - 40),
+                            cv2.FONT_HERSHEY_SIMPLEX, 0.6, (0, 0, 255), 2)
+            else:
+                cv2.putText(frame, "Waiting to Lock", (x1, y1 - 40),
+                            cv2.FONT_HERSHEY_SIMPLEX, 0.6, (255, 255, 0), 2)
+            
+            if x1 > 600 and x1 < 780:
+                if final_label == "DefectiveLemon":
+                    mqtt_message = f"lemon_classification classification=\"{final_label}\" {int(time.time()*1e9)}"
+                    if time.time() - lemon_time > .3:
+                        mqtt_client.publish(MQTT_TOPIC, mqtt_message)
+                        lemon_time = time.time()
+                        i = i + 1
+                elif final_label == "NotRipeLemon":
+                    mqtt_message = f"lemon_classification classification=\"{final_label}\" {int(time.time()*1e9)}"
+                    if time.time() - lemon_time > .3:
+                        mqtt_client.publish(MQTT_TOPIC, mqtt_message)
+                        lemon_time = time.time()
+                        i = i + 1
+                elif final_label == "GoodLemon":
+                    mqtt_message = f"lemon_classification classification=\"{final_label}\" {int(time.time()*1e9)}"
+                    if time.time() - lemon_time > .3:
+                        mqtt_client.publish(MQTT_TOPIC, mqtt_message)
+                        lemon_time = time.time()
+                        i = i + 1
 
     # Display the processed video stream
     cv2.imshow("Live Detection", frame)

Test_Track_updated.py

@@ -1,8 +1,40 @@
 import cv2
 from ultralytics import YOLO
 from collections import deque
+import paho.mqtt.client as mqtt
+from influxdb import InfluxDBClient
+from influxdb_client import InfluxDBClient, Point, WriteOptions
+import time
+from datetime import datetime
+import ssl
 import os
 
+# InfluxDB Configuration
+INFLUX_URL = "http://localhost:8086"
+INFLUX_TOKEN = "export INFLUX_TOKEN=duVTQHPpHqr6WmdYfpSStqm-pxnvZHs-W0-3lXDnk8Tn6PGt59MlnTSR6egjMWdYvmL_ZI6xt3YUzGVBZHvc7w=="
+INFLUX_ORG = "GAAIM"
+INFLUX_BUCKET = "AGVIGNETTE"
+
+# Connect to InfluxDB
+client = InfluxDBClient(url=INFLUX_URL, token=INFLUX_TOKEN, org=INFLUX_ORG)
+write_api = client.write_api(write_options=WriteOptions(batch_size=1))
+
+# MQTT Setup
+MQTT_BROKER = "192.168.10.57"
+MQTT_TOPIC = "fruit/classification"
+
+mqtt_client = mqtt.Client()
+
+
+# Set up TLS/SSL for MQTT connection
+
+# mqtt_client.tls_set(
+#    ca_certs="/Users/vel/Desktop/CvModel/mosquitto/mosquitto/certs/ca.crt",  # Path to the CA certificate
+#    tls_version=ssl.PROTOCOL_TLS  # Specify the TLS version
+#)
+#mqtt_client.tls_insecure_set(True)
+mqtt_client.connect(MQTT_BROKER, 1883, 6000)
+
 # Allow duplicate loading of OpenMP runtime
 os.environ["KMP_DUPLICATE_LIB_OK"] = "True"
 
@@ -10,10 +42,15 @@ os.environ["KMP_DUPLICATE_LIB_OK"] = "True"
 yaml_path = "botsort.yaml"
 
 # Camera index (default camera index, 1 indicates an external camera)
-camera_index = 1
+camera_index = 0
+
+cap = cv2.VideoCapture(camera_index)
+cap.set(cv2.CAP_PROP_FPS, 30)
+width = int(cap.get(cv2.CAP_PROP_FRAME_WIDTH))
+height = int(cap.get(cv2.CAP_PROP_FRAME_HEIGHT))
 
 # Load the YOLO model
-model = YOLO(r"D:\AIM\lemon\runs\detect\train4\weights\best.pt")  # Load custom model
+model = YOLO(r"/Users/vel/Desktop/CvModel/CV_AG/runs/detect/train4/weights/best.pt")  # Load custom model
 
 # Define class labels
 class_labels = {
@@ -34,6 +71,7 @@ GOOD_THRESHOLD = 0.7    # Threshold for "GoodLemon" and "NotRipeLemon" proportio
 
 # State history for each target (used for smoothing), format: {ID: deque([...], maxlen=HISTORY_LENGTH)}
 lemon_history = {}
+lemon_send_history = []
 
 # Set the display window to be resizable
 cv2.namedWindow("Live Detection", cv2.WINDOW_NORMAL)
@@ -71,15 +109,17 @@ def get_smoothed_label(obj_id, current_label):
 # Use streaming tracking mode to maintain tracker state
 results = model.track(
     source=camera_index,    # Get video stream directly from the camera
-    conf=0.5,
+    conf=0.45,
     tracker=yaml_path,      # Use the YAML configuration file
     persist=True,           # Persist tracking (do not reset)
     stream=True,            # Stream processing, not frame-by-frame calling
-    show=False
+    show=False,
+    device = 'mps' #'cpu' 
 )
 
 # Iterate over streaming tracking results
 for result in results:
+
     frame = result.orig_img  # Current frame
     detections = result.boxes  # Detection box information
 
@@ -100,6 +140,32 @@ for result in results:
                 if final_label in smoothing_labels:
                     position = f"({x1}, {y1}, {x2}, {y2})"
                     print(f"ID: {obj_id}, Position: {position}, Label: {display_text}")
+                            # Draw detection box and label with color based on classification
+                    if final_label == "DefectiveLemon":
+                        box_color = (100, 100, 255)  # Red for defective
+                    elif final_label == "NotRipeLemon":
+                        box_color = (255, 100, 80)  # Blue for unripe
+                    elif final_label == "GoodLemon":
+                        box_color = (0, 255, 0)  # Green for good
+                    else:
+                        box_color = (255, 255, 255)  # White for unknown or other classes
+
+                    # Add background rectangle for text
+                    text_size = cv2.getTextSize(display_text, cv2.FONT_HERSHEY_TRIPLEX, 0.6, 2)[0]
+                    text_x, text_y = x1, y1 - 10
+                    text_w, text_h = text_size[0], text_size[1]
+                    cv2.rectangle(frame, (text_x, text_y - text_h - 5), (text_x + text_w, text_y + 5), (0, 0, 0), -1)
+
+                    # Draw detection box and text
+                    cv2.rectangle(frame, (x1, y1), (x2, y2), box_color, 2)
+                    cv2.putText(frame, display_text, (text_x, text_y),
+                                cv2.FONT_HERSHEY_TRIPLEX, 0.6, box_color, 2)
+
+                    if x1 > 750 and x1 < 850 and y2 < 410 and y1 > 190 and obj_id not in lemon_send_history:
+                        if final_label in ["DefectiveLemon", "NotRipeLemon", "GoodLemon"]:
+                            mqtt_message = f"lemon_classification classification=\"{final_label}\" {int(time.time()*1e9)}"
+                            lemon_send_history.append(obj_id)
+                            mqtt_client.publish(MQTT_TOPIC, mqtt_message)
             else:
                 # For other classes, display the current detection result directly and clear history (if exists)
                 if obj_id in lemon_history:
@@ -108,11 +174,6 @@ for result in results:
         else:
             display_text = label
 
-        # Draw detection box and label
-        cv2.rectangle(frame, (x1, y1), (x2, y2), (0, 255, 0), 2)
-        cv2.putText(frame, display_text, (x1, y1 - 10),
-                    cv2.FONT_HERSHEY_SIMPLEX, 0.6, (0, 255, 0), 2)
-
     # Display the processed frame
     cv2.imshow("Live Detection", frame)
 
@@ -123,3 +184,5 @@ for result in results:
 
 cv2.destroyAllWindows()
 print("Camera video processing complete. Program terminated.")
+
+

botsort.yaml

@@ -4,18 +4,18 @@
 # For documentation and examples see https://docs.ultralytics.com/modes/track/
 # For BoT-SORT source code see https://github.com/NirAharon/BoT-SORT
 
-tracker_type: botsort # tracker type, ['botsort', 'bytetrack']
+tracker_type: bytetrack # tracker type, ['botsort', 'bytetrack']
-track_high_thresh: 0.25 # threshold for the first association
+track_high_thresh: 0.2 # threshold for the first association
-track_low_thresh: 0.1 # threshold for the second association
+track_low_thresh: 0.05 # threshold for the second association
-new_track_thresh: 0.4 # threshold for init new track if the detection does not match any tracks
+new_track_thresh: 0.3 # threshold for init new track if the detection does not match any tracks
-track_buffer: 30 # buffer to calculate the time when to remove tracks
+track_buffer: 50 # buffer to calculate the time when to remove tracks
-match_thresh: 0.7 # threshold for matching tracks
+match_thresh: 0.8 # threshold for matching tracks
 fuse_score: True # Whether to fuse confidence scores with the iou distances before matching
 # min_box_area: 10  # threshold for min box areas(for tracker evaluation, not used for now)
 
 # BoT-SORT settings
 gmc_method: sparseOptFlow # method of global motion compensation
 # ReID model related thresh (not supported yet)
-proximity_thresh: 0.5
+proximity_thresh: 0.6
-appearance_thresh: 0.25
+appearance_thresh: 0.2
 with_reid: False