File size: 5,794 Bytes
050c87a
 
 
 
 
 
 
 
bf269bc
050c87a
 
 
 
 
c54e670
050c87a
 
 
e89b7ca
050c87a
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
a129d63
 
050c87a
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
161120c
d89de93
161120c
 
 
 
 
 
 
 
 
 
 
 
e90d845
161120c
a129d63
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
161120c
 
 
 
 
 
d89de93
 
 
 
 
050c87a
c15f918
050c87a
 
 
 
 
 
c77a11e
7cdb8cc
c54e670
 
 
8ae2da0
161120c
c77a11e
075a84b
 
292393c
c54e670
8ae2da0
161120c
f633603
 
 
 
 
 
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
import matplotlib.pyplot as plt
import numpy as np
from six import BytesIO
from PIL import Image
import tensorflow as tf
from object_detection.utils import label_map_util
from object_detection.utils import visualization_utils as viz_utils
from object_detection.utils import ops as utils_op
from tqdm import tqdm
import tarfile
import wget 
import gradio as gr
from huggingface_hub import snapshot_download
import os 
import cv2

PATH_TO_LABELS = 'data/label_map.pbtxt'   
category_index = label_map_util.create_category_index_from_labelmap(PATH_TO_LABELS, use_display_name=True)
label_id_offset = 0

def pil_image_as_numpy_array(pilimg):

    img_array = tf.keras.utils.img_to_array(pilimg)
    img_array = np.expand_dims(img_array, axis=0)
    return img_array
    
def load_image_into_numpy_array(path):
                                    
    image = None
    image_data = tf.io.gfile.GFile(path, 'rb').read()
    image = Image.open(BytesIO(image_data))
    return pil_image_as_numpy_array(image)            

def load_model():
    download_dir = snapshot_download(REPO_ID)
    saved_model_dir = os.path.join(download_dir, "saved_model")
    detection_model = tf.saved_model.load(saved_model_dir)
    return detection_model

def load_model2():
    wget.download("https://nyp-aicourse.s3-ap-southeast-1.amazonaws.com/pretrained-models/balloon_model.tar.gz")
    tarfile.open("balloon_model.tar.gz").extractall()
    model_dir = 'saved_model'    
    detection_model = tf.saved_model.load(str(model_dir))
    return detection_model    

# samples_folder = 'test_samples
# image_path = 'test_samples/sample_balloon.jpeg
# 

def predict(pilimg):
    image_np = pil_image_as_numpy_array(pilimg)
    return predict2(image_np)  

def predict2(image_np):

    results = detection_model(image_np)

    # different object detection models have additional results
    result = {key:value.numpy() for key,value in results.items()}
    
    label_id_offset = 0
    image_np_with_detections = image_np.copy()

    viz_utils.visualize_boxes_and_labels_on_image_array(
        image_np_with_detections[0],
        result['detection_boxes'][0],
        (result['detection_classes'][0] + label_id_offset).astype(int),
        result['detection_scores'][0],
        category_index,
        use_normalized_coordinates=True,
        max_boxes_to_draw=200,
        min_score_thresh=.60,
        agnostic_mode=False,
        line_thickness=2)

    result_pil_img = tf.keras.utils.array_to_img(image_np_with_detections[0])
    
    return result_pil_img
def write_video(video_in_filepath, video_out_filepath, detection_model):
       
    video_reader = cv2.VideoCapture(video_in_filepath)
    
    nb_frames = int(video_reader.get(cv2.CAP_PROP_FRAME_COUNT))
    frame_h = int(video_reader.get(cv2.CAP_PROP_FRAME_HEIGHT))
    frame_w = int(video_reader.get(cv2.CAP_PROP_FRAME_WIDTH))
    fps = video_reader.get(cv2.CAP_PROP_FPS)
    
    video_writer = cv2.VideoWriter(video_out_filepath,
                               cv2.VideoWriter_fourcc(*'mp4v'), 
                               fps, 
                               (frame_w, frame_h))

    while True:
        ret, image_np = video_reader.read()
        if not ret:
            break

        results = predict(image_np)
        results_np = np.array(results)
        video_writer.write(results_np)
        #input_tensor = tf.convert_to_tensor(np.expand_dims(image_np, 0), dtype=tf.uint8)
        #results = detection_model(input_tensor)
        #viz_utils.visualize_boxes_and_labels_on_image_array(
        #          image_np,
        #         results['detection_boxes'][0].numpy(),
        #          (results['detection_classes'][0].numpy()+ label_id_offset).astype(int),
        #          results['detection_scores'][0].numpy(),
        #          category_index,
        #          use_normalized_coordinates=True,
        #          max_boxes_to_draw=200,
        #          min_score_thresh=.50,
        #          agnostic_mode=False,
        #          line_thickness=2)

        #video_writer.write(np.uint8(image_np))
                
    # Release camera and close windows
    video_reader.release()
    video_writer.release() 
    cv2.destroyAllWindows() 
    cv2.waitKey(1)
    
def predict_video (video_file_name):
    detected_video_file = "detected_video.mp4"
    write_video(video_file_name,detected_video_file,detection_model)
    return detected_video_file

REPO_ID = "YEHTUT/tfodmodel"
detection_model = load_model()
# pil_image = Image.open(image_path)
# image_arr = pil_image_as_numpy_array(pil_image)

# predicted_img = predict(image_arr)
# predicted_img.save('predicted.jpg')
Image_tab = gr.Interface(fn=predict,
             inputs=gr.Image(type="pil"),
             outputs=gr.Image(type="pil"),
             examples=[["SampleImage1.jpg"],["SampleImage2.jpg"],["SampleImage3.jpg"],["SampleImage4.jpg"],["SampleImage5.jpg"],["SampleImage6.jpg"]],
            title="This is the object detection model for Durian and Pineapple images",
            description="Using ssd_mobilenet_v2_320x320_coco17_tpu-8 to detect Durian and Pineapple"
             )
Video_tab = gr.Interface(fn=predict_video,
             inputs=gr.Video(label="Upload Video"),
             outputs=gr.Video(label="Detected Video"),
             examples=[["SampleVideo1.mp4"],["SampleVideo2.mp4"]],
            title="This is the object detection model for Durian and Pineapple videos",
            description="Using ssd_mobilenet_v2_320x320_coco17_tpu-8 to detect Durian and Pineapple"                         
             )

gr.TabbedInterface([Image_tab, Video_tab], ["Image", "Video"]).launch(share=True)
#gr.Interface(fn=predict,
#             inputs=gr.Image(type="pil"),
#             outputs=gr.Image(type="pil")
#             ).launch(share=True)