File size: 7,076 Bytes
c4a1c35
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
# -*- coding: utf-8 -*-
"""Judol Gradio YOLO11.ipynb

Automatically generated by Colab.

Original file is located at
    https://colab.research.google.com/drive/1oiuTAi-cys1ydtUhSDJSRdeA02mAmZQH
"""



import cv2
from ultralytics import YOLO
import gradio as gr
import imageio




model = YOLO('https://huggingface.co/JrEasy/Judol-Detection-YOLO11/resolve/main/best.pt')


confidence_threshold = 0.6

class_names = {
    0: "BK8",
    1: "Gate of Olympus",
    2: "Princess",
    3: "Starlight Princess",
    4: "Zeus",
}

class_colors = {
    0: (0, 255, 0),       # Green for BK8
    1: (255, 0, 0),       # Blue for Gate of Olympus
    2: (0, 0, 255),       # Red for Princess
    3: (255, 255, 0),     # Cyan for Starlight Princess
    4: (255, 0, 255),     # Magenta for Zeus
}

def format_time_ranges(timestamps, classes):

    if not timestamps:
        return ""


    class_timestamps = {}

    for timestamp, class_id in zip(timestamps, classes):
        class_name = class_names.get(class_id, 'Unknown')
        if class_name not in class_timestamps:
            class_timestamps[class_name] = []
        class_timestamps[class_name].append(timestamp)


    formatted_ranges = []

    for class_name, timestamps in class_timestamps.items():
        timestamps = sorted(timestamps)
        ranges = []
        start = timestamps[0]
        for i in range(1, len(timestamps)):
            if timestamps[i] - timestamps[i - 1] <= 1:
                continue
            else:
                ranges.append(f"{int(start)}-{int(timestamps[i - 1])}")
                start = timestamps[i]

        ranges.append(f"{int(start)}-{int(timestamps[-1])}")

        formatted_ranges.append(f"{class_name} = {', '.join(ranges)}")

    return ", ".join(formatted_ranges)

import os

def process_video(input_video):
    cap = cv2.VideoCapture(input_video)
    if not cap.isOpened():
        print("Error: Could not open input video.")
        return None, []

    fps = cap.get(cv2.CAP_PROP_FPS)
    frame_width = int(cap.get(cv2.CAP_PROP_FRAME_WIDTH))
    frame_height = int(cap.get(cv2.CAP_PROP_FRAME_HEIGHT))

    # Define the output video path in the current directory
    output_video_path = os.path.join(os.getcwd(), "processed_video.mp4")

    writer = imageio.get_writer(output_video_path, fps=fps, codec="h264")

    frame_count = 0
    timestamps = []
    classes_detected = []

    while cap.isOpened():
        ret, frame = cap.read()
        if not ret:
            break

        timestamp = frame_count / fps
        frame_count += 1

        # Resize the frame to 640x640 before passing to the model
        resized_frame = cv2.resize(frame, (640, 640))

        gray_frame = cv2.cvtColor(resized_frame, cv2.COLOR_BGR2GRAY)
        input_frame = cv2.merge([gray_frame, gray_frame, gray_frame])

        results = model.predict(input_frame)

        for result in results:
            for box in result.boxes:
                if box.conf[0] >= confidence_threshold:
                    x1, y1, x2, y2 = map(int, box.xyxy[0])
                    class_id = int(box.cls[0])
                    class_name = class_names.get(class_id, f"Class {class_id}")
                    color = class_colors.get(class_id, (0, 255, 0))
                    cv2.rectangle(resized_frame, (x1, y1), (x2, y2), color, 2)
                    text = f'{class_name}, Conf: {box.conf[0]:.2f}'
                    text_position = (x1, y1 - 10 if y1 > 20 else y1 + 20)
                    cv2.putText(resized_frame, text, text_position, cv2.FONT_HERSHEY_SIMPLEX, 0.5, color, 2)

                    timestamps.append(timestamp)
                    classes_detected.append(class_id)

        # Resize the frame back to original size for the output video
        output_frame = cv2.resize(resized_frame, (frame_width, frame_height))

        writer.append_data(cv2.cvtColor(output_frame, cv2.COLOR_BGR2RGB))

    cap.release()
    writer.close()

    formatted_time_ranges = format_time_ranges(timestamps, classes_detected)

    print(f"Processed video saved at: {output_video_path}")

    return output_video_path, formatted_time_ranges




def process_image(input_image):
    # Convert image from RGB to BGR for OpenCV processing
    bgr_frame = cv2.cvtColor(input_image, cv2.COLOR_RGB2BGR)

    # Resize the frame to 640x640 before passing to the model
    resized_frame = cv2.resize(bgr_frame, (640, 640))

    # Convert to grayscale and create a 3-channel grayscale image
    gray_frame = cv2.cvtColor(resized_frame, cv2.COLOR_BGR2GRAY)
    input_frame = cv2.merge([gray_frame, gray_frame, gray_frame])

    
    results = model.predict(input_frame)

    detections_log = []  
    classes_detected = []  

    for result in results:
        for box in result.boxes:
            if box.conf[0] >= confidence_threshold:  
                x1, y1, x2, y2 = map(int, box.xyxy[0]) 
                class_id = int(box.cls[0]) 
                class_name = class_names.get(class_id, f"Class {class_id}")
                color = class_colors.get(class_id, (0, 255, 0))  # Default green color

               
                cv2.rectangle(resized_frame, (x1, y1), (x2, y2), color, 2)
                text = f'{class_name}, Conf: {box.conf[0]:.2f}'
                text_position = (x1, y1 - 10 if y1 > 20 else y1 + 20)
                cv2.putText(resized_frame, text, text_position, cv2.FONT_HERSHEY_SIMPLEX, 0.5, color, 2)

               
                detections_log.append({
                    "class": class_name,
                    "confidence": box.conf[0]
                })
                classes_detected.append(class_id)

    # Count occurrences of each class detected
    class_count = {class_names.get(cls, f"Class {cls}"): classes_detected.count(cls) for cls in set(classes_detected)}

    # Format the detections as 'Class = Count' pairs
    formatted_log = ", ".join([f"{class_name} = {count}" for class_name, count in class_count.items()])

    # Convert the output frame back to RGB
    output_image = cv2.cvtColor(resized_frame, cv2.COLOR_BGR2RGB)
    return output_image, formatted_log


with gr.Blocks() as app:
    gr.Markdown("## Judol Detection using YOLOv11")

    with gr.Tab("Video Detection"):
        with gr.Row():
            input_video = gr.Video(label="Upload a video")
            output_video = gr.Video(label="Processed Video")
            detections_log = gr.Textbox(label="Detections Log", lines=10)

        input_video.change(
            fn=lambda input_video: process_video(input_video) if input_video else ("", []),
            inputs=input_video,
            outputs=[output_video, detections_log],
        )

    with gr.Tab("Image Detection"):
        with gr.Row():
            input_image = gr.Image(label="Upload an image")
            output_image = gr.Image(label="Processed Image")
            image_detections_log = gr.Textbox(label="Detections Log", lines=10)

        input_image.change(
            fn=process_image,
            inputs=input_image,
            outputs=[output_image, image_detections_log],
        )
        
app.launch()