refactor: single predict.py entrypoint (production + CLI), archive experiments

archive/diagnose_truck_jetson.py

@@ -0,0 +1,94 @@
+import sys
+import os
+import cv2
+import torch
+import time
+import numpy as np
+from ultralytics import YOLO
+from shapely.geometry import Polygon, Point
+
+def main():
+    print("=== Detailed Jetson Detection Test ===")
+    
+    # 1. Load Zones
+    zones_path = "/home/jetson/karung/zones.json"
+    truck_pts = []
+    if os.path.exists(zones_path):
+        import json
+        with open(zones_path, 'r') as f:
+            data = json.load(f)
+            truck_pts = data.get('truck', [])
+            print(f"Zones.json truck points: {truck_pts}")
+    
+    # Target resolution
+    target_w, target_h = 1280, 720
+    
+    # Scale factors assuming zones were drawn on 1920x1080
+    orig_w, orig_h = 1920, 1080
+    scale_x = target_w / orig_w
+    scale_y = target_h / orig_h
+    
+    scaled_truck_pts = [[int(p[0] * scale_x), int(p[1] * scale_y)] for p in truck_pts] if truck_pts else [
+        [389, 294], [398, 718], [885, 719], [885, 277]
+    ]
+    truck_polygon = Polygon(scaled_truck_pts)
+    print(f"Scaled truck polygon: {scaled_truck_pts}")
+    
+    # 2. Open Stream
+    source = "rtsp://192.168.192.96:8554/cam"
+    print(f"Connecting to RTSP stream: {source}...")
+    cap = cv2.VideoCapture(source)
+    if not cap.isOpened():
+        print("Error: Could not open RTSP source.")
+        return
+        
+    # Wait for the stream to warm up and buffer
+    print("Warming up stream reader for 3 seconds...")
+    time.sleep(3.0)
+    
+    # 3. Load Model
+    model_path = "/home/jetson/karung/model_karung_truk.engine"
+    print(f"Loading TensorRT Model: {model_path}...")
+    model = YOLO(model_path)
+    
+    print("Running 10 frames of inference...")
+    detections_summary = {}
+    
+    frame_count = 0
+    attempts = 0
+    while frame_count < 10 and attempts < 100:
+        ret, frame = cap.read()
+        attempts += 1
+        if not ret or frame is None:
+            time.sleep(0.1)
+            continue
+            
+        frame_count += 1
+        frame_resized = cv2.resize(frame, (target_w, target_h))
+        results = model(frame_resized, conf=0.01, imgsz=640, device="cuda", verbose=False)
+        result = results[0]
+        
+        detected_in_frame = []
+        for box in result.boxes:
+            cls_id = int(box.cls[0])
+            name = model.names[cls_id]
+            conf = float(box.conf[0])
+            x1, y1, x2, y2 = box.xyxy[0].tolist()
+            tcx = (x1 + x2) / 2.0
+            tcy = (y1 + y2) / 2.0
+            
+            # Check if inside truck polygon
+            pt = Point(tcx, tcy)
+            in_poly = truck_polygon.contains(pt)
+            
+            detected_in_frame.append(f"{name} ({conf:.3f}) at ({tcx:.1f},{tcy:.1f}) in_poly={in_poly}")
+            detections_summary[name] = detections_summary.get(name, 0) + 1
+            
+        print(f"Frame {frame_count} (attempt {attempts}): {', '.join(detected_in_frame) if detected_in_frame else 'None'}")
+            
+    cap.release()
+    print("\nSummary of detected objects over processed frames:")
+    print(detections_summary)
+
+if __name__ == "__main__":
+    main()