Update app.py
Browse files
app.py
CHANGED
@@ -6,156 +6,147 @@ import gradio as gr
|
|
6 |
from scipy.interpolate import interp1d
|
7 |
import uuid
|
8 |
import os
|
|
|
9 |
|
10 |
-
# Load the trained YOLOv8n model
|
11 |
-
model = YOLO("best.pt")
|
12 |
|
13 |
-
# Constants
|
14 |
-
STUMPS_WIDTH = 0.2286 # meters
|
15 |
-
BALL_DIAMETER = 0.073 # meters
|
16 |
FRAME_RATE = 30 # Input video frame rate
|
17 |
-
SLOW_MOTION_FACTOR =
|
18 |
-
CONF_THRESHOLD = 0.3
|
|
|
|
|
|
|
19 |
|
20 |
def process_video(video_path):
|
21 |
-
|
22 |
if not os.path.exists(video_path):
|
23 |
return [], [], "Error: Video file not found"
|
24 |
cap = cv2.VideoCapture(video_path)
|
25 |
frames = []
|
26 |
ball_positions = []
|
27 |
debug_log = []
|
28 |
-
|
29 |
frame_count = 0
|
|
|
|
|
30 |
while cap.isOpened():
|
31 |
ret, frame = cap.read()
|
32 |
if not ret:
|
33 |
break
|
34 |
frame_count += 1
|
|
|
|
|
|
|
|
|
|
|
35 |
frames.append(frame.copy()) # Store original frame
|
36 |
-
# Detect ball
|
37 |
-
results = model.predict(
|
38 |
detections = 0
|
39 |
for detection in results[0].boxes:
|
40 |
-
if detection.cls == 0: #
|
41 |
detections += 1
|
42 |
x1, y1, x2, y2 = detection.xyxy[0].cpu().numpy()
|
|
|
|
|
43 |
ball_positions.append([(x1 + x2) / 2, (y1 + y2) / 2])
|
44 |
-
# Draw bounding box on frame
|
45 |
cv2.rectangle(frame, (int(x1), int(y1)), (int(x2), int(y2)), (0, 255, 0), 2)
|
46 |
-
frames[-1] = frame
|
47 |
debug_log.append(f"Frame {frame_count}: {detections} ball detections")
|
|
|
|
|
|
|
48 |
cap.release()
|
49 |
|
|
|
50 |
if not ball_positions:
|
51 |
debug_log.append("No balls detected in any frame")
|
52 |
else:
|
53 |
debug_log.append(f"Total ball detections: {len(ball_positions)}")
|
54 |
-
|
55 |
return frames, ball_positions, "\n".join(debug_log)
|
56 |
|
57 |
def estimate_trajectory(ball_positions, frames):
|
58 |
-
|
59 |
if len(ball_positions) < 2:
|
60 |
return None, None, "Error: Fewer than 2 ball detections for trajectory"
|
61 |
-
# Extract x, y coordinates
|
62 |
x_coords = [pos[0] for pos in ball_positions]
|
63 |
y_coords = [pos[1] for pos in ball_positions]
|
64 |
times = np.arange(len(ball_positions)) / FRAME_RATE
|
65 |
-
|
66 |
-
# Interpolate to smooth trajectory
|
67 |
try:
|
68 |
fx = interp1d(times, x_coords, kind='linear', fill_value="extrapolate")
|
69 |
fy = interp1d(times, y_coords, kind='quadratic', fill_value="extrapolate")
|
70 |
except Exception as e:
|
71 |
return None, None, f"Error in trajectory interpolation: {str(e)}"
|
72 |
-
|
73 |
-
# Project trajectory forward (0.5 seconds post-impact)
|
74 |
t_future = np.linspace(times[-1], times[-1] + 0.5, 10)
|
75 |
x_future = fx(t_future)
|
76 |
y_future = fy(t_future)
|
77 |
-
|
78 |
-
return list(zip(x_future, y_future)), t_future,
|
79 |
|
80 |
def lbw_decision(ball_positions, trajectory, frames):
|
81 |
-
|
82 |
if not frames:
|
83 |
return "Error: No frames processed", None, None, None
|
84 |
if not trajectory or len(ball_positions) < 2:
|
85 |
return "Not enough data (insufficient ball detections)", None, None, None
|
86 |
-
|
87 |
-
# Assume stumps are at the bottom center of the frame (calibration needed)
|
88 |
frame_height, frame_width = frames[0].shape[:2]
|
89 |
stumps_x = frame_width / 2
|
90 |
-
stumps_y = frame_height * 0.9
|
91 |
-
stumps_width_pixels = frame_width * (STUMPS_WIDTH / 3.0)
|
92 |
-
|
93 |
-
# Store pitch and impact points
|
94 |
pitch_point = ball_positions[0]
|
95 |
impact_point = ball_positions[-1]
|
96 |
-
|
97 |
-
# Check pitching point
|
98 |
pitch_x, pitch_y = pitch_point
|
99 |
if pitch_x < stumps_x - stumps_width_pixels / 2 or pitch_x > stumps_x + stumps_width_pixels / 2:
|
100 |
return f"Not Out (Pitched outside line at x: {pitch_x:.1f}, y: {pitch_y:.1f})", trajectory, pitch_point, impact_point
|
101 |
-
|
102 |
-
# Check impact point
|
103 |
impact_x, impact_y = impact_point
|
104 |
if impact_x < stumps_x - stumps_width_pixels / 2 or impact_x > stumps_x + stumps_width_pixels / 2:
|
105 |
return f"Not Out (Impact outside line at x: {impact_x:.1f}, y: {impact_y:.1f})", trajectory, pitch_point, impact_point
|
106 |
-
|
107 |
-
# Check trajectory hitting stumps
|
108 |
for x, y in trajectory:
|
109 |
if abs(x - stumps_x) < stumps_width_pixels / 2 and abs(y - stumps_y) < frame_height * 0.1:
|
110 |
return f"Out (Ball hits stumps, Pitch at x: {pitch_x:.1f}, y: {pitch_y:.1f}, Impact at x: {impact_x:.1f}, y: {impact_y:.1f})", trajectory, pitch_point, impact_point
|
|
|
111 |
return f"Not Out (Missing stumps, Pitch at x: {pitch_x:.1f}, y: {pitch_y:.1f}, Impact at x: {impact_x:.1f}, y: {impact_y:.1f})", trajectory, pitch_point, impact_point
|
112 |
|
113 |
def generate_slow_motion(frames, trajectory, pitch_point, impact_point, output_path):
|
114 |
-
|
115 |
if not frames:
|
116 |
return None
|
117 |
fourcc = cv2.VideoWriter_fourcc(*'mp4v')
|
118 |
out = cv2.VideoWriter(output_path, fourcc, FRAME_RATE / SLOW_MOTION_FACTOR, (frames[0].shape[1], frames[0].shape[0]))
|
119 |
-
|
120 |
for frame in frames:
|
121 |
-
# Draw trajectory
|
122 |
if trajectory:
|
123 |
for x, y in trajectory:
|
124 |
-
cv2.circle(frame, (int(x), int(y)), 5, (255, 0, 0), -1)
|
125 |
-
|
126 |
-
# Draw pitch point (red circle with label)
|
127 |
if pitch_point:
|
128 |
x, y = pitch_point
|
129 |
-
cv2.circle(frame, (int(x), int(y)), 8, (0, 0, 255), -1)
|
130 |
cv2.putText(frame, "Pitch Point", (int(x) + 10, int(y) - 10),
|
131 |
cv2.FONT_HERSHEY_SIMPLEX, 0.6, (0, 0, 255), 2)
|
132 |
-
|
133 |
-
# Draw impact point (yellow circle with label)
|
134 |
if impact_point:
|
135 |
x, y = impact_point
|
136 |
-
cv2.circle(frame, (int(x), int(y)), 8, (0, 255, 255), -1)
|
137 |
cv2.putText(frame, "Impact Point", (int(x) + 10, int(y) + 20),
|
138 |
cv2.FONT_HERSHEY_SIMPLEX, 0.6, (0, 255, 255), 2)
|
139 |
-
|
140 |
-
for _ in range(SLOW_MOTION_FACTOR): # Duplicate frames for very slow motion
|
141 |
out.write(frame)
|
142 |
out.release()
|
143 |
-
|
|
|
144 |
|
145 |
def drs_review(video):
|
146 |
-
|
147 |
frames, ball_positions, debug_log = process_video(video)
|
148 |
if not frames:
|
149 |
return f"Error: Failed to process video\nDebug Log:\n{debug_log}", None
|
150 |
trajectory, _, trajectory_log = estimate_trajectory(ball_positions, frames)
|
151 |
decision, trajectory, pitch_point, impact_point = lbw_decision(ball_positions, trajectory, frames)
|
152 |
-
|
153 |
-
# Generate slow-motion replay with enhanced annotations
|
154 |
output_path = f"output_{uuid.uuid4()}.mp4"
|
155 |
-
slow_motion_path = generate_slow_motion(frames, trajectory, pitch_point, impact_point, output_path)
|
156 |
-
|
157 |
-
# Combine debug logs for output
|
158 |
-
debug_output = f"{debug_log}\n{trajectory_log}"
|
159 |
return f"DRS Decision: {decision}\nDebug Log:\n{debug_output}", slow_motion_path
|
160 |
|
161 |
# Gradio interface
|
@@ -164,10 +155,10 @@ iface = gr.Interface(
|
|
164 |
inputs=gr.Video(label="Upload Video Clip"),
|
165 |
outputs=[
|
166 |
gr.Textbox(label="DRS Decision and Debug Log"),
|
167 |
-
gr.Video(label="
|
168 |
],
|
169 |
title="AI-Powered DRS for LBW in Local Cricket",
|
170 |
-
description="Upload a video clip of a cricket delivery to get an LBW decision and
|
171 |
)
|
172 |
|
173 |
if __name__ == "__main__":
|
|
|
6 |
from scipy.interpolate import interp1d
|
7 |
import uuid
|
8 |
import os
|
9 |
+
import time
|
10 |
|
11 |
+
# Load the trained YOLOv8n model
|
12 |
+
model = YOLO("best.pt")
|
13 |
|
14 |
+
# Constants
|
15 |
+
STUMPS_WIDTH = 0.2286 # meters
|
16 |
+
BALL_DIAMETER = 0.073 # meters
|
17 |
FRAME_RATE = 30 # Input video frame rate
|
18 |
+
SLOW_MOTION_FACTOR = 3 # Reduced from 6 for faster processing
|
19 |
+
CONF_THRESHOLD = 0.3
|
20 |
+
MAX_DETECTIONS = 10 # Stop after detecting enough ball positions
|
21 |
+
PROCESS_EVERY_N_FRAME = 2 # Process every 2nd frame
|
22 |
+
RESIZE_FACTOR = 0.5 # Downscale frames to 50% for faster processing
|
23 |
|
24 |
def process_video(video_path):
|
25 |
+
start_time = time.time()
|
26 |
if not os.path.exists(video_path):
|
27 |
return [], [], "Error: Video file not found"
|
28 |
cap = cv2.VideoCapture(video_path)
|
29 |
frames = []
|
30 |
ball_positions = []
|
31 |
debug_log = []
|
|
|
32 |
frame_count = 0
|
33 |
+
processed_frames = 0
|
34 |
+
|
35 |
while cap.isOpened():
|
36 |
ret, frame = cap.read()
|
37 |
if not ret:
|
38 |
break
|
39 |
frame_count += 1
|
40 |
+
if frame_count % PROCESS_EVERY_N_FRAME != 0:
|
41 |
+
continue # Skip frames
|
42 |
+
processed_frames += 1
|
43 |
+
# Resize frame for faster processing
|
44 |
+
frame_small = cv2.resize(frame, (0, 0), fx=RESIZE_FACTOR, fy=RESIZE_FACTOR)
|
45 |
frames.append(frame.copy()) # Store original frame
|
46 |
+
# Detect ball
|
47 |
+
results = model.predict(frame_small, conf=CONF_THRESHOLD)
|
48 |
detections = 0
|
49 |
for detection in results[0].boxes:
|
50 |
+
if detection.cls == 0: # Ball class
|
51 |
detections += 1
|
52 |
x1, y1, x2, y2 = detection.xyxy[0].cpu().numpy()
|
53 |
+
# Scale coordinates back to original frame size
|
54 |
+
x1, y1, x2, y2 = [v / RESIZE_FACTOR for v in [x1, y1, x2, y2]]
|
55 |
ball_positions.append([(x1 + x2) / 2, (y1 + y2) / 2])
|
56 |
+
# Draw bounding box on original frame
|
57 |
cv2.rectangle(frame, (int(x1), int(y1)), (int(x2), int(y2)), (0, 255, 0), 2)
|
58 |
+
frames[-1] = frame
|
59 |
debug_log.append(f"Frame {frame_count}: {detections} ball detections")
|
60 |
+
if len(ball_positions) >= MAX_DETECTIONS:
|
61 |
+
debug_log.append(f"Stopping early after {len(ball_positions)} detections")
|
62 |
+
break
|
63 |
cap.release()
|
64 |
|
65 |
+
debug_log.append(f"Processed {processed_frames} frames in {time.time() - start_time:.2f} seconds")
|
66 |
if not ball_positions:
|
67 |
debug_log.append("No balls detected in any frame")
|
68 |
else:
|
69 |
debug_log.append(f"Total ball detections: {len(ball_positions)}")
|
|
|
70 |
return frames, ball_positions, "\n".join(debug_log)
|
71 |
|
72 |
def estimate_trajectory(ball_positions, frames):
|
73 |
+
start_time = time.time()
|
74 |
if len(ball_positions) < 2:
|
75 |
return None, None, "Error: Fewer than 2 ball detections for trajectory"
|
|
|
76 |
x_coords = [pos[0] for pos in ball_positions]
|
77 |
y_coords = [pos[1] for pos in ball_positions]
|
78 |
times = np.arange(len(ball_positions)) / FRAME_RATE
|
|
|
|
|
79 |
try:
|
80 |
fx = interp1d(times, x_coords, kind='linear', fill_value="extrapolate")
|
81 |
fy = interp1d(times, y_coords, kind='quadratic', fill_value="extrapolate")
|
82 |
except Exception as e:
|
83 |
return None, None, f"Error in trajectory interpolation: {str(e)}"
|
|
|
|
|
84 |
t_future = np.linspace(times[-1], times[-1] + 0.5, 10)
|
85 |
x_future = fx(t_future)
|
86 |
y_future = fy(t_future)
|
87 |
+
debug_log = f"Trajectory estimated in {time.time() - start_time:.2f} seconds"
|
88 |
+
return list(zip(x_future, y_future)), t_future, debug_log
|
89 |
|
90 |
def lbw_decision(ball_positions, trajectory, frames):
|
91 |
+
start_time = time.time()
|
92 |
if not frames:
|
93 |
return "Error: No frames processed", None, None, None
|
94 |
if not trajectory or len(ball_positions) < 2:
|
95 |
return "Not enough data (insufficient ball detections)", None, None, None
|
|
|
|
|
96 |
frame_height, frame_width = frames[0].shape[:2]
|
97 |
stumps_x = frame_width / 2
|
98 |
+
stumps_y = frame_height * 0.9
|
99 |
+
stumps_width_pixels = frame_width * (STUMPS_WIDTH / 3.0)
|
|
|
|
|
100 |
pitch_point = ball_positions[0]
|
101 |
impact_point = ball_positions[-1]
|
|
|
|
|
102 |
pitch_x, pitch_y = pitch_point
|
103 |
if pitch_x < stumps_x - stumps_width_pixels / 2 or pitch_x > stumps_x + stumps_width_pixels / 2:
|
104 |
return f"Not Out (Pitched outside line at x: {pitch_x:.1f}, y: {pitch_y:.1f})", trajectory, pitch_point, impact_point
|
|
|
|
|
105 |
impact_x, impact_y = impact_point
|
106 |
if impact_x < stumps_x - stumps_width_pixels / 2 or impact_x > stumps_x + stumps_width_pixels / 2:
|
107 |
return f"Not Out (Impact outside line at x: {impact_x:.1f}, y: {impact_y:.1f})", trajectory, pitch_point, impact_point
|
|
|
|
|
108 |
for x, y in trajectory:
|
109 |
if abs(x - stumps_x) < stumps_width_pixels / 2 and abs(y - stumps_y) < frame_height * 0.1:
|
110 |
return f"Out (Ball hits stumps, Pitch at x: {pitch_x:.1f}, y: {pitch_y:.1f}, Impact at x: {impact_x:.1f}, y: {impact_y:.1f})", trajectory, pitch_point, impact_point
|
111 |
+
debug_log = f"LBW decision computed in {time.time() - start_time:.2f} seconds"
|
112 |
return f"Not Out (Missing stumps, Pitch at x: {pitch_x:.1f}, y: {pitch_y:.1f}, Impact at x: {impact_x:.1f}, y: {impact_y:.1f})", trajectory, pitch_point, impact_point
|
113 |
|
114 |
def generate_slow_motion(frames, trajectory, pitch_point, impact_point, output_path):
|
115 |
+
start_time = time.time()
|
116 |
if not frames:
|
117 |
return None
|
118 |
fourcc = cv2.VideoWriter_fourcc(*'mp4v')
|
119 |
out = cv2.VideoWriter(output_path, fourcc, FRAME_RATE / SLOW_MOTION_FACTOR, (frames[0].shape[1], frames[0].shape[0]))
|
|
|
120 |
for frame in frames:
|
|
|
121 |
if trajectory:
|
122 |
for x, y in trajectory:
|
123 |
+
cv2.circle(frame, (int(x), int(y)), 5, (255, 0, 0), -1)
|
|
|
|
|
124 |
if pitch_point:
|
125 |
x, y = pitch_point
|
126 |
+
cv2.circle(frame, (int(x), int(y)), 8, (0, 0, 255), -1)
|
127 |
cv2.putText(frame, "Pitch Point", (int(x) + 10, int(y) - 10),
|
128 |
cv2.FONT_HERSHEY_SIMPLEX, 0.6, (0, 0, 255), 2)
|
|
|
|
|
129 |
if impact_point:
|
130 |
x, y = impact_point
|
131 |
+
cv2.circle(frame, (int(x), int(y)), 8, (0, 255, 255), -1)
|
132 |
cv2.putText(frame, "Impact Point", (int(x) + 10, int(y) + 20),
|
133 |
cv2.FONT_HERSHEY_SIMPLEX, 0.6, (0, 255, 255), 2)
|
134 |
+
for _ in range(SLOW_MOTION_FACTOR):
|
|
|
135 |
out.write(frame)
|
136 |
out.release()
|
137 |
+
debug_log = f"Slow-motion video generated in {time.time() - start_time:.2f} seconds"
|
138 |
+
return output_path, debug_log
|
139 |
|
140 |
def drs_review(video):
|
141 |
+
start_time = time.time()
|
142 |
frames, ball_positions, debug_log = process_video(video)
|
143 |
if not frames:
|
144 |
return f"Error: Failed to process video\nDebug Log:\n{debug_log}", None
|
145 |
trajectory, _, trajectory_log = estimate_trajectory(ball_positions, frames)
|
146 |
decision, trajectory, pitch_point, impact_point = lbw_decision(ball_positions, trajectory, frames)
|
|
|
|
|
147 |
output_path = f"output_{uuid.uuid4()}.mp4"
|
148 |
+
slow_motion_path, slow_motion_log = generate_slow_motion(frames, trajectory, pitch_point, impact_point, output_path)
|
149 |
+
debug_output = f"{debug_log}\n{trajectory_log}\n{slow_motion_log}\nTotal processing time: {time.time() - start_time:.2f} seconds"
|
|
|
|
|
150 |
return f"DRS Decision: {decision}\nDebug Log:\n{debug_output}", slow_motion_path
|
151 |
|
152 |
# Gradio interface
|
|
|
155 |
inputs=gr.Video(label="Upload Video Clip"),
|
156 |
outputs=[
|
157 |
gr.Textbox(label="DRS Decision and Debug Log"),
|
158 |
+
gr.Video(label="Slow-Motion Replay with Ball Detection (Green), Trajectory (Blue), Pitch Point (Red), Impact Point (Yellow)")
|
159 |
],
|
160 |
title="AI-Powered DRS for LBW in Local Cricket",
|
161 |
+
description="Upload a 3-second video clip of a cricket delivery to get an LBW decision and slow-motion replay showing ball detection (green boxes), trajectory (blue dots), pitch point (red circle), and impact point (yellow circle)."
|
162 |
)
|
163 |
|
164 |
if __name__ == "__main__":
|