Create app.py
Browse files
app.py
ADDED
@@ -0,0 +1,80 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import gradio as gr
|
2 |
+
import torch
|
3 |
+
import cv2
|
4 |
+
import numpy as np
|
5 |
+
import matplotlib.pyplot as plt
|
6 |
+
from yolov5 import YOLOv5
|
7 |
+
|
8 |
+
# Load YOLOv5 model (best.pt)
|
9 |
+
model = YOLOv5("best.pt") # Adjust the path to your model file
|
10 |
+
|
11 |
+
# Function to process the video and calculate ball trajectory, speed, and visualize the pitch
|
12 |
+
def process_video(video_file):
|
13 |
+
# Load video file using OpenCV
|
14 |
+
video = cv2.VideoCapture(video_file.name)
|
15 |
+
ball_positions = []
|
16 |
+
speed_data = []
|
17 |
+
|
18 |
+
frame_count = 0
|
19 |
+
last_position = None
|
20 |
+
|
21 |
+
while video.isOpened():
|
22 |
+
ret, frame = video.read()
|
23 |
+
if not ret:
|
24 |
+
break
|
25 |
+
|
26 |
+
frame_count += 1
|
27 |
+
|
28 |
+
# Run YOLOv5 model on the frame to detect ball
|
29 |
+
results = model(frame)
|
30 |
+
|
31 |
+
# Extract the ball position (assuming class 0 = ball)
|
32 |
+
ball_detections = results.pandas().xywh
|
33 |
+
ball = ball_detections[ball_detections['class'] == 0] # class 0 is ball, adjust as needed
|
34 |
+
|
35 |
+
if not ball.empty:
|
36 |
+
ball_x = ball.iloc[0]['xmin'] + (ball.iloc[0]['xmax'] - ball.iloc[0]['xmin']) / 2
|
37 |
+
ball_y = ball.iloc[0]['ymin'] + (ball.iloc[0]['ymax'] - ball.iloc[0]['ymin']) / 2
|
38 |
+
ball_positions.append((frame_count, ball_x, ball_y)) # Track position in each frame
|
39 |
+
|
40 |
+
if last_position is not None:
|
41 |
+
# Calculate speed based on pixel displacement between frames
|
42 |
+
distance = np.sqrt((ball_x - last_position[1]) ** 2 + (ball_y - last_position[2]) ** 2)
|
43 |
+
fps = video.get(cv2.CAP_PROP_FPS) # Frames per second of the video
|
44 |
+
speed = distance * fps # Speed = distance / time (time between frames is 1/fps)
|
45 |
+
speed_data.append(speed)
|
46 |
+
|
47 |
+
last_position = (frame_count, ball_x, ball_y) # Update last position
|
48 |
+
|
49 |
+
video.release()
|
50 |
+
|
51 |
+
# Ball trajectory plot
|
52 |
+
plot_trajectory(ball_positions)
|
53 |
+
|
54 |
+
# Return results
|
55 |
+
avg_speed = np.mean(speed_data) if speed_data else 0
|
56 |
+
return f"Average Ball Speed: {avg_speed:.2f} pixels per second"
|
57 |
+
|
58 |
+
# Function to plot ball trajectory using matplotlib
|
59 |
+
def plot_trajectory(ball_positions):
|
60 |
+
x_positions = [pos[1] for pos in ball_positions]
|
61 |
+
y_positions = [pos[2] for pos in ball_positions]
|
62 |
+
|
63 |
+
plt.figure(figsize=(10, 6))
|
64 |
+
plt.plot(x_positions, y_positions, label="Ball Trajectory", color='b')
|
65 |
+
plt.title("Ball Trajectory on Pitch")
|
66 |
+
plt.xlabel("X Position (pitch width)")
|
67 |
+
plt.ylabel("Y Position (pitch length)")
|
68 |
+
plt.grid(True)
|
69 |
+
plt.legend()
|
70 |
+
plt.show()
|
71 |
+
|
72 |
+
# Gradio interface for the app
|
73 |
+
iface = gr.Interface(
|
74 |
+
fn=process_video, # Function to call when video is uploaded
|
75 |
+
inputs=gr.inputs.File(label="Upload a Video File"), # File input (video)
|
76 |
+
outputs="text", # Output the result as text
|
77 |
+
live=True # Keep the interface live
|
78 |
+
)
|
79 |
+
|
80 |
+
iface.launch(debug=True)
|