Spaces:
Runtime error
Runtime error
File size: 5,035 Bytes
200d507 94e7125 200d507 |
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 |
import gradio as gr
from gradio_client import Client , handle_file
import cv2
import os
import shutil
from PIL import Image
from moviepy import *
clientImgPipeLn = Client("dj-dawgs-ipd/IPD-Image-Pipeline")
clientAudioPipeLn = Client("dj-dawgs-ipd/IPD-Audio-Pipeline")
def predict(video_path):
cap = cv2.VideoCapture(video_path)
fps = int(cap.get(cv2.CAP_PROP_FPS))
frame_interval = fps * 2
frame_count = 0
success = True
temp_data_path = "temp_data"
os.makedirs(temp_data_path, exist_ok=True)
temp_frames_path = os.path.join(temp_data_path, "temp_frames")
os.makedirs(temp_frames_path, exist_ok=True)
resImg = {}
resAudio = {}
video_clip = VideoFileClip(video_path)
if video_clip.audio is None:
resAudio = {
'prediction' : None,
'language' : None,
'label' : None,
'confidence' : None,
'hate_text' : None
}
else:
audio_path = os.path.join(temp_data_path , "temp_audio.wav")
video_clip.audio.write_audiofile(audio_path , codec="pcm_s16le")
resAudio = clientAudioPipeLn.predict(
audio_path = handle_file(audio_path),
api_name = '/predict'
)
while success:
success, frame = cap.read()
if frame_count % frame_interval == 0 and success:
temp_image_path = os.path.join(temp_data_path, f"temp_frames/temp_frame_{frame_count // fps}s.jpg")
cv2.imwrite(temp_image_path, frame)
response = clientImgPipeLn.predict(
image=handle_file(temp_image_path),
api_name="/predict"
)
print(f"Response for frame at {frame_count // fps}s: {response}")
if response['prediction'] == 'hate':
resImg = response
resImg['hate_image_timestamp'] = frame_count//fps
break
frame_count += 1
cap.release()
shutil.rmtree(temp_data_path)
if len(resImg) == 0 and resAudio['prediction'] == 'not_hate':
return {
'prediction' : 'not_hate',
'language' : {
'video' : None,
'audio' : None
},
'label' : {
'video' : None,
'audio' : None
},
'confidence' : None,
'hate_text' : {
'video' : None,
'audio' : None
},
'hate_image_timestamp' : None,
'hate_component' : None
}
if resImg['prediction'] == 'hate' and resAudio['prediction'] == 'not_hate':
resImg['hate_component'] = 'video'
return {
'prediction' : 'hate',
'language' : {
'video' : resImg['language'],
'audio' : None
},
'label' : {
'video' : resImg['label'],
'audio' : None
},
'confidence' : resImg['confidence'],
'hate_text' : {
'video' : resImg['hate_text'],
'audio' : None
},
'hate_image_timestamp' : resImg['hate_image_timestamp'],
'hate_component' : ["video"]
}
if len(resImg) == 0 and resAudio['prediction'] == 'hate':
return {
'prediction' : 'hate',
'language' : {
'video' : None,
'audio' : resAudio['language']
},
'label' : {
'video' : None,
'audio' : resAudio['label']
},
'confidence' : resAudio['confidence'],
'hate_text' : {
'video' : None,
'audio' : resAudio['hate_text']
},
'hate_image_timestamp' : None,
'hate_component' : ["audio"],
}
return {
'prediction' : 'hate',
'language' : {
'video' : resImg['language'],
'audio' : resAudio['language']
},
'label' : {
'video' : resImg['label'],
'audio' : resAudio['label']
},
'confidence' : ((resImg['confidence'] or 0) + (resAudio['confidence'] or 0)) / (2 - (resImg['confidence'] == None or resAudio['confidence'] == None)),
'hate_text' : {
'video' : resImg['hate_text'],
'audio' : resAudio['hate_text']
},
'hate_image_timestamp' : resImg['hate_image_timestamp'],
'hate_component' : ["video" , "audio"]
}
iface = gr.Interface(fn=predict,
inputs = gr.Video(),
outputs=gr.JSON(),
title = "Hate Speech Detection in Video",
description = "Detect hateful symbols or text in Video"
)
if __name__ == "__main__":
iface.launch(show_error = True)
|