Update app.py
Browse files
app.py
CHANGED
@@ -15,15 +15,17 @@ STUMPS_WIDTH = 0.2286 # meters (width of stumps)
|
|
15 |
BALL_DIAMETER = 0.073 # meters (approx. cricket ball diameter)
|
16 |
FRAME_RATE = 30 # Input video frame rate
|
17 |
SLOW_MOTION_FACTOR = 6 # For very slow motion (6x slower)
|
18 |
-
CONF_THRESHOLD = 0.25 #
|
19 |
IMPACT_ZONE_Y = 0.85 # Fraction of frame height where impact is likely (near stumps)
|
|
|
20 |
|
21 |
def process_video(video_path):
|
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
|
@@ -40,6 +42,7 @@ def process_video(video_path):
|
|
40 |
detections += 1
|
41 |
x1, y1, x2, y2 = detection.xyxy[0].cpu().numpy()
|
42 |
ball_positions.append([(x1 + x2) / 2, (y1 + y2) / 2])
|
|
|
43 |
cv2.rectangle(frame, (int(x1), int(y1)), (int(x2), int(y2)), (0, 255, 0), 2)
|
44 |
frames[-1] = frame
|
45 |
debug_log.append(f"Frame {frame_count}: {detections} ball detections")
|
@@ -50,48 +53,34 @@ def process_video(video_path):
|
|
50 |
else:
|
51 |
debug_log.append(f"Total ball detections: {len(ball_positions)}")
|
52 |
|
53 |
-
|
54 |
-
if ball_positions:
|
55 |
-
ball_positions = interpolate_missing_positions(ball_positions, frame_count)
|
56 |
|
57 |
-
|
58 |
-
|
59 |
-
def interpolate_missing_positions(ball_positions, total_frames):
|
60 |
-
if len(ball_positions) < 2:
|
61 |
-
return ball_positions
|
62 |
-
times = np.linspace(0, total_frames / FRAME_RATE, total_frames)
|
63 |
-
detected_times = [i / FRAME_RATE for i, _ in enumerate(ball_positions)]
|
64 |
-
x_coords = [pos[0] for pos in ball_positions]
|
65 |
-
y_coords = [pos[1] for pos in ball_positions]
|
66 |
-
|
67 |
-
try:
|
68 |
-
fx = interp1d(detected_times, x_coords, kind='linear', fill_value="extrapolate")
|
69 |
-
fy = interp1d(detected_times, y_coords, kind='linear', fill_value="extrapolate")
|
70 |
-
interpolated_positions = [(fx(t), fy(t)) for t in times if t <= detected_times[-1]]
|
71 |
-
return interpolated_positions
|
72 |
-
except:
|
73 |
-
return ball_positions
|
74 |
-
|
75 |
-
def estimate_trajectory(ball_positions, frames):
|
76 |
if len(ball_positions) < 2:
|
77 |
-
return None, None, None, "Error: Fewer than 2 ball detections for trajectory"
|
78 |
frame_height = frames[0].shape[0]
|
79 |
|
80 |
# Extract x, y coordinates
|
81 |
x_coords = [pos[0] for pos in ball_positions]
|
82 |
y_coords = [pos[1] for pos in ball_positions]
|
83 |
-
times = np.
|
|
|
|
|
|
|
|
|
84 |
|
85 |
-
# Find impact point (
|
86 |
impact_idx = None
|
87 |
-
|
88 |
-
|
|
|
89 |
impact_idx = i
|
|
|
90 |
break
|
91 |
if impact_idx is None:
|
92 |
-
impact_idx = len(ball_positions) - 1
|
|
|
93 |
|
94 |
-
pitch_point = ball_positions[0]
|
95 |
impact_point = ball_positions[impact_idx]
|
96 |
|
97 |
# Use positions up to impact for interpolation
|
@@ -103,15 +92,16 @@ def estimate_trajectory(ball_positions, frames):
|
|
103 |
fx = interp1d(times, x_coords, kind='linear', fill_value="extrapolate")
|
104 |
fy = interp1d(times, y_coords, kind='quadratic', fill_value="extrapolate")
|
105 |
except Exception as e:
|
106 |
-
return None, None, None, f"Error in trajectory interpolation: {str(e)}"
|
107 |
|
108 |
-
# Project trajectory (detected + future)
|
109 |
t_full = np.linspace(times[0], times[-1] + 0.5, len(times) + 10)
|
110 |
x_full = fx(t_full)
|
111 |
y_full = fy(t_full)
|
112 |
trajectory = list(zip(x_full, y_full))
|
113 |
|
114 |
-
|
|
|
115 |
|
116 |
def lbw_decision(ball_positions, trajectory, frames, pitch_point, impact_point):
|
117 |
if not frames:
|
@@ -141,31 +131,33 @@ def lbw_decision(ball_positions, trajectory, frames, pitch_point, impact_point):
|
|
141 |
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
|
142 |
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
|
143 |
|
144 |
-
def generate_slow_motion(frames, trajectory, pitch_point, impact_point, output_path):
|
145 |
if not frames:
|
146 |
return None
|
147 |
fourcc = cv2.VideoWriter_fourcc(*'mp4v')
|
148 |
out = cv2.VideoWriter(output_path, fourcc, FRAME_RATE / SLOW_MOTION_FACTOR, (frames[0].shape[1], frames[0].shape[0]))
|
149 |
|
150 |
-
|
151 |
-
|
152 |
-
|
153 |
-
|
154 |
-
|
|
|
|
|
155 |
|
156 |
-
# Draw pitch point (red circle with label)
|
157 |
-
if pitch_point:
|
158 |
x, y = pitch_point
|
159 |
cv2.circle(frame, (int(x), int(y)), 8, (0, 0, 255), -1)
|
160 |
cv2.putText(frame, "Pitch Point", (int(x) + 10, int(y) - 10),
|
161 |
-
|
162 |
|
163 |
-
# Draw impact point (yellow circle with label)
|
164 |
-
if impact_point:
|
165 |
x, y = impact_point
|
166 |
cv2.circle(frame, (int(x), int(y)), 8, (0, 255, 255), -1)
|
167 |
cv2.putText(frame, "Impact Point", (int(x) + 10, int(y) + 20),
|
168 |
-
|
169 |
|
170 |
for _ in range(SLOW_MOTION_FACTOR):
|
171 |
out.write(frame)
|
@@ -173,14 +165,14 @@ def generate_slow_motion(frames, trajectory, pitch_point, impact_point, output_p
|
|
173 |
return output_path
|
174 |
|
175 |
def drs_review(video):
|
176 |
-
frames, ball_positions, debug_log = process_video(video)
|
177 |
if not frames:
|
178 |
return f"Error: Failed to process video\nDebug Log:\n{debug_log}", None
|
179 |
-
trajectory, pitch_point, impact_point, trajectory_log = estimate_trajectory(ball_positions, frames)
|
180 |
decision, trajectory, pitch_point, impact_point = lbw_decision(ball_positions, trajectory, frames, pitch_point, impact_point)
|
181 |
|
182 |
output_path = f"output_{uuid.uuid4()}.mp4"
|
183 |
-
slow_motion_path = generate_slow_motion(frames, trajectory, pitch_point, impact_point, output_path)
|
184 |
|
185 |
debug_output = f"{debug_log}\n{trajectory_log}"
|
186 |
return f"DRS Decision: {decision}\nDebug Log:\n{debug_output}", slow_motion_path
|
@@ -191,10 +183,10 @@ iface = gr.Interface(
|
|
191 |
inputs=gr.Video(label="Upload Video Clip"),
|
192 |
outputs=[
|
193 |
gr.Textbox(label="DRS Decision and Debug Log"),
|
194 |
-
gr.Video(label="Very Slow-Motion Replay with Ball Detection (Green), Trajectory (Blue), Pitch Point (Red), Impact Point (Yellow)")
|
195 |
],
|
196 |
title="AI-Powered DRS for LBW in Local Cricket",
|
197 |
-
description="Upload a video clip of a cricket delivery to get an LBW decision and very slow-motion replay showing ball detection (green boxes), trajectory (blue
|
198 |
)
|
199 |
|
200 |
if __name__ == "__main__":
|
|
|
15 |
BALL_DIAMETER = 0.073 # meters (approx. cricket ball diameter)
|
16 |
FRAME_RATE = 30 # Input video frame rate
|
17 |
SLOW_MOTION_FACTOR = 6 # For very slow motion (6x slower)
|
18 |
+
CONF_THRESHOLD = 0.25 # Confidence threshold for detection
|
19 |
IMPACT_ZONE_Y = 0.85 # Fraction of frame height where impact is likely (near stumps)
|
20 |
+
IMPACT_DELTA_Y = 50 # Pixels for detecting sudden y-position change (impact)
|
21 |
|
22 |
def process_video(video_path):
|
23 |
if not os.path.exists(video_path):
|
24 |
+
return [], [], [], "Error: Video file not found"
|
25 |
cap = cv2.VideoCapture(video_path)
|
26 |
frames = []
|
27 |
ball_positions = []
|
28 |
+
detection_frames = [] # Track frames with detections
|
29 |
debug_log = []
|
30 |
|
31 |
frame_count = 0
|
|
|
42 |
detections += 1
|
43 |
x1, y1, x2, y2 = detection.xyxy[0].cpu().numpy()
|
44 |
ball_positions.append([(x1 + x2) / 2, (y1 + y2) / 2])
|
45 |
+
detection_frames.append(frame_count - 1) # Store frame index (0-based)
|
46 |
cv2.rectangle(frame, (int(x1), int(y1)), (int(x2), int(y2)), (0, 255, 0), 2)
|
47 |
frames[-1] = frame
|
48 |
debug_log.append(f"Frame {frame_count}: {detections} ball detections")
|
|
|
53 |
else:
|
54 |
debug_log.append(f"Total ball detections: {len(ball_positions)}")
|
55 |
|
56 |
+
return frames, ball_positions, detection_frames, "\n".join(debug_log)
|
|
|
|
|
57 |
|
58 |
+
def estimate_trajectory(ball_positions, frames, detection_frames):
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
59 |
if len(ball_positions) < 2:
|
60 |
+
return None, None, None, None, "Error: Fewer than 2 ball detections for trajectory"
|
61 |
frame_height = frames[0].shape[0]
|
62 |
|
63 |
# Extract x, y coordinates
|
64 |
x_coords = [pos[0] for pos in ball_positions]
|
65 |
y_coords = [pos[1] for pos in ball_positions]
|
66 |
+
times = np.array(detection_frames) / FRAME_RATE
|
67 |
+
|
68 |
+
# Identify pitch point (first detection)
|
69 |
+
pitch_point = ball_positions[0]
|
70 |
+
pitch_frame = detection_frames[0]
|
71 |
|
72 |
+
# Find impact point (sudden change in y or near stumps)
|
73 |
impact_idx = None
|
74 |
+
impact_frame = None
|
75 |
+
for i in range(1, len(y_coords)):
|
76 |
+
if y_coords[i] > frame_height * IMPACT_ZONE_Y or abs(y_coords[i] - y_coords[i-1]) > IMPACT_DELTA_Y:
|
77 |
impact_idx = i
|
78 |
+
impact_frame = detection_frames[i]
|
79 |
break
|
80 |
if impact_idx is None:
|
81 |
+
impact_idx = len(ball_positions) - 1
|
82 |
+
impact_frame = detection_frames[-1]
|
83 |
|
|
|
84 |
impact_point = ball_positions[impact_idx]
|
85 |
|
86 |
# Use positions up to impact for interpolation
|
|
|
92 |
fx = interp1d(times, x_coords, kind='linear', fill_value="extrapolate")
|
93 |
fy = interp1d(times, y_coords, kind='quadratic', fill_value="extrapolate")
|
94 |
except Exception as e:
|
95 |
+
return None, None, None, None, f"Error in trajectory interpolation: {str(e)}"
|
96 |
|
97 |
+
# Project trajectory (detected + future for LBW decision)
|
98 |
t_full = np.linspace(times[0], times[-1] + 0.5, len(times) + 10)
|
99 |
x_full = fx(t_full)
|
100 |
y_full = fy(t_full)
|
101 |
trajectory = list(zip(x_full, y_full))
|
102 |
|
103 |
+
debug_log = f"Trajectory estimated successfully\nPitch point at frame {pitch_frame + 1}: ({pitch_point[0]:.1f}, {pitch_point[1]:.1f})\nImpact point at frame {impact_frame + 1}: ({impact_point[0]:.1f}, {impact_point[1]:.1f})"
|
104 |
+
return trajectory, pitch_point, impact_point, pitch_frame, impact_frame, debug_log
|
105 |
|
106 |
def lbw_decision(ball_positions, trajectory, frames, pitch_point, impact_point):
|
107 |
if not frames:
|
|
|
131 |
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
|
132 |
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
|
133 |
|
134 |
+
def generate_slow_motion(frames, trajectory, pitch_point, impact_point, detection_frames, pitch_frame, impact_frame, output_path):
|
135 |
if not frames:
|
136 |
return None
|
137 |
fourcc = cv2.VideoWriter_fourcc(*'mp4v')
|
138 |
out = cv2.VideoWriter(output_path, fourcc, FRAME_RATE / SLOW_MOTION_FACTOR, (frames[0].shape[1], frames[0].shape[0]))
|
139 |
|
140 |
+
# Extract trajectory points up to impact for visualization
|
141 |
+
trajectory_points = np.array(trajectory[:len(detection_frames)], dtype=np.int32).reshape((-1, 1, 2))
|
142 |
+
|
143 |
+
for i, frame in enumerate(frames):
|
144 |
+
# Draw trajectory (blue line) only for frames with detections
|
145 |
+
if i in detection_frames and trajectory_points.size > 0:
|
146 |
+
cv2.polylines(frame, [trajectory_points[:detection_frames.index(i) + 1]], False, (255, 0, 0), 2)
|
147 |
|
148 |
+
# Draw pitch point (red circle with label) only in pitch frame
|
149 |
+
if pitch_point and i == pitch_frame:
|
150 |
x, y = pitch_point
|
151 |
cv2.circle(frame, (int(x), int(y)), 8, (0, 0, 255), -1)
|
152 |
cv2.putText(frame, "Pitch Point", (int(x) + 10, int(y) - 10),
|
153 |
+
cv2.FONT_HERSHEY_SIMPLEX, 0.6, (0, 0, 255), 2)
|
154 |
|
155 |
+
# Draw impact point (yellow circle with label) only in impact frame
|
156 |
+
if impact_point and i == impact_frame:
|
157 |
x, y = impact_point
|
158 |
cv2.circle(frame, (int(x), int(y)), 8, (0, 255, 255), -1)
|
159 |
cv2.putText(frame, "Impact Point", (int(x) + 10, int(y) + 20),
|
160 |
+
cv2.FONT_HERSHEY_SIMPLEX, 0.6, (0, 255, 255), 2)
|
161 |
|
162 |
for _ in range(SLOW_MOTION_FACTOR):
|
163 |
out.write(frame)
|
|
|
165 |
return output_path
|
166 |
|
167 |
def drs_review(video):
|
168 |
+
frames, ball_positions, detection_frames, debug_log = process_video(video)
|
169 |
if not frames:
|
170 |
return f"Error: Failed to process video\nDebug Log:\n{debug_log}", None
|
171 |
+
trajectory, pitch_point, impact_point, pitch_frame, impact_frame, trajectory_log = estimate_trajectory(ball_positions, frames, detection_frames)
|
172 |
decision, trajectory, pitch_point, impact_point = lbw_decision(ball_positions, trajectory, frames, pitch_point, impact_point)
|
173 |
|
174 |
output_path = f"output_{uuid.uuid4()}.mp4"
|
175 |
+
slow_motion_path = generate_slow_motion(frames, trajectory, pitch_point, impact_point, detection_frames, pitch_frame, impact_frame, output_path)
|
176 |
|
177 |
debug_output = f"{debug_log}\n{trajectory_log}"
|
178 |
return f"DRS Decision: {decision}\nDebug Log:\n{debug_output}", slow_motion_path
|
|
|
183 |
inputs=gr.Video(label="Upload Video Clip"),
|
184 |
outputs=[
|
185 |
gr.Textbox(label="DRS Decision and Debug Log"),
|
186 |
+
gr.Video(label="Very Slow-Motion Replay with Ball Detection (Green), Trajectory (Blue Line), Pitch Point (Red), Impact Point (Yellow)")
|
187 |
],
|
188 |
title="AI-Powered DRS for LBW in Local Cricket",
|
189 |
+
description="Upload a video clip of a cricket delivery to get an LBW decision and very slow-motion replay showing ball detection (green boxes), trajectory (blue line), pitch point (red circle), and impact point (yellow circle)."
|
190 |
)
|
191 |
|
192 |
if __name__ == "__main__":
|