wip

.gitignore

@@ -0,0 +1,2 @@
+yolo.h5
+venv

README.md

@@ -0,0 +1 @@
+download yolo model https://github.com/OlafenwaMoses/ImageAI/releases/tag/1.0/

main.py

@@ -0,0 +1,94 @@
+from imageai.Detection import VideoObjectDetection
+import matplotlib.pyplot as plt
+
+
+vid_obj_detect = VideoObjectDetection()
+vid_obj_detect.setModelTypeAsYOLOv3()
+
+vid_obj_detect.setModelPath(r"/home/mikolaj/2ait_tech/sportowe/film/yolo.h5")
+vid_obj_detect.loadModel()
+
+
+person_points = []
+ball_points = []
+
+
+def forFrame(frame_number, output_array, output_count, detected_frame):
+    for index, object in enumerate(output_array):
+        if object['name'] == 'sports ball':
+            ball_points.append((index, object['box_points'][0], object['box_points'][1]))
+        else:
+            person_points.append((index, object['box_points'][0], object['box_points'][1]))
+
+detected_vid_obj = vid_obj_detect.detectObjectsFromVideo(
+    input_file_path =  r"football.mp4",
+    output_file_path = r"changed_football.mp4",
+    frames_per_second=20,
+    log_progress=True,
+    per_frame_function = forFrame,
+    # per_second_function=forSeconds,
+    return_detected_frame = True,
+)
+
+with open('saved_cordinats', 'w') as file:
+    file.write(str(person_points))
+    file.write(str(ball_points))
+
+
+fig = plt.figure()
+ax = fig.add_subplot(111)
+# person_points = [(0, 206, 376), (1, 1166, 405), (2, 42, 362), (3, 710, 390), (4, 754, 408), (5, 1083, 378), (0, 216, 383), (1, 64, 367), (2, 709, 387), (3, 750, 400), (4, 1087, 374), (5, 1179, 389)]
+# ball_points= [(6, 204, 564), (6, 211, 566), (6, 225, 563)]
+
+COLORS = ['red', 'blue', 'orange', 'pink', 'yellow', 'brown']
+LABELS = ['p1', 'p2', 'p3', 'p4', 'p5', 'p6']
+
+persons = {
+    0:{
+        'x': [],
+        'y': []
+    },
+    1:{
+        'x': [],
+        'y': []
+    },
+    2:{
+        'x': [],
+        'y': []
+    },
+    3:{
+        'x': [],
+        'y': []
+    },
+    4:{
+        'x': [],
+        'y': []
+    },
+    5:{
+        'x': [],
+        'y': []
+    },
+    # 6:{
+    #     'x': [],
+    #     'y': []
+    # },
+}
+
+try:
+
+    for index, p in enumerate(person_points):
+        persons[p[0]]['x'].append(p[1])
+        persons[p[0]]['y'].append(p[2])
+    for k, v in persons.items():
+        ax.plot( persons[k]['x'], persons[k]['y'], color=COLORS[k], lw=2, label=LABELS[k]) 
+    
+    ball_x = [x[1] for x in ball_points]
+    ball_y = [x[2] for x in ball_points]
+    ax.plot(ball_x, ball_y, color='green', lw=2, label='ball')
+except:
+    pass
+
+ax.legend()
+
+plt.savefig('wykres.png')
+plt.show()

req.txt

@@ -0,0 +1,6 @@
+imageAI==2.1.6
+opencv-python
+Pillow
+numpy==1.21
+tensorflow
+torch torchvision torchaudio --extra-index-url https://download.pytorch.org/whl/cpu