Spaces:
Runtime error
Runtime error
Update app.py
Browse files
app.py
CHANGED
|
@@ -1,39 +1,41 @@
|
|
| 1 |
import torch
|
| 2 |
-
import torch.nn as nn
|
| 3 |
import gradio as gr
|
| 4 |
import cv2
|
| 5 |
import numpy as np
|
| 6 |
|
| 7 |
-
#
|
| 8 |
-
|
|
|
|
|
|
|
|
|
|
| 9 |
def __init__(self):
|
| 10 |
super(YourModelArchitecture, self).__init__()
|
| 11 |
-
#
|
| 12 |
-
# Example: self.conv1 = nn.Conv2d(3, 16, kernel_size=3, stride=1, padding=1)
|
| 13 |
-
|
| 14 |
-
def forward(self, x):
|
| 15 |
-
# Define the forward pass logic
|
| 16 |
-
return x
|
| 17 |
|
| 18 |
-
|
| 19 |
-
|
|
|
|
| 20 |
|
| 21 |
-
#
|
| 22 |
def load_model(model_path):
|
| 23 |
-
|
| 24 |
-
model
|
| 25 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
| 26 |
return model
|
| 27 |
|
| 28 |
-
#
|
| 29 |
def preprocess_frame(frame):
|
| 30 |
-
#
|
| 31 |
-
frame = cv2.resize(frame, (224, 224)) #
|
| 32 |
-
frame = frame / 255.0 #
|
| 33 |
-
input_tensor = torch.from_numpy(frame.astype(np.float32)).permute(2, 0, 1) #
|
| 34 |
-
return input_tensor.unsqueeze(0) #
|
| 35 |
|
| 36 |
-
#
|
| 37 |
def process_video(model, video_path):
|
| 38 |
cap = cv2.VideoCapture(video_path)
|
| 39 |
fourcc = cv2.VideoWriter_fourcc(*'mp4v')
|
|
@@ -45,36 +47,36 @@ def process_video(model, video_path):
|
|
| 45 |
if not ret:
|
| 46 |
break
|
| 47 |
|
| 48 |
-
#
|
| 49 |
input_tensor = preprocess_frame(frame)
|
| 50 |
|
| 51 |
-
#
|
| 52 |
with torch.no_grad():
|
| 53 |
predictions = model(input_tensor)
|
| 54 |
|
| 55 |
-
#
|
| 56 |
output_frame = (predictions.squeeze().permute(1, 2, 0).numpy() * 255).astype(np.uint8)
|
| 57 |
|
| 58 |
-
#
|
| 59 |
out.write(output_frame)
|
| 60 |
|
| 61 |
cap.release()
|
| 62 |
out.release()
|
| 63 |
return output_path
|
| 64 |
|
| 65 |
-
# Gradio
|
| 66 |
def colorize_video(video):
|
| 67 |
model = load_model(MODEL_PATH)
|
| 68 |
-
output_video_path = process_video(model, video.name) #
|
| 69 |
return output_video_path
|
| 70 |
|
| 71 |
-
#
|
| 72 |
iface = gr.Interface(
|
| 73 |
fn=colorize_video,
|
| 74 |
-
inputs=gr.Video(label="
|
| 75 |
-
outputs=gr.Video(label="
|
| 76 |
-
title="
|
| 77 |
-
description="
|
| 78 |
)
|
| 79 |
|
| 80 |
if __name__ == '__main__':
|
|
|
|
| 1 |
import torch
|
|
|
|
| 2 |
import gradio as gr
|
| 3 |
import cv2
|
| 4 |
import numpy as np
|
| 5 |
|
| 6 |
+
# Chemin vers le modèle
|
| 7 |
+
MODEL_PATH = 'ColorizeVideo_gen.pth'
|
| 8 |
+
|
| 9 |
+
# Définir l'architecture de votre modèle ici
|
| 10 |
+
class YourModelArchitecture(torch.nn.Module):
|
| 11 |
def __init__(self):
|
| 12 |
super(YourModelArchitecture, self).__init__()
|
| 13 |
+
# Définissez les couches de votre modèle
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 14 |
|
| 15 |
+
def forward(self, x):
|
| 16 |
+
# Définissez la logique de passage avant de votre modèle
|
| 17 |
+
return x # Modifiez ceci selon votre modèle
|
| 18 |
|
| 19 |
+
# Charger le modèle
|
| 20 |
def load_model(model_path):
|
| 21 |
+
checkpoint = torch.load(model_path, map_location=torch.device('cpu')) # Charger le checkpoint
|
| 22 |
+
model = YourModelArchitecture() # Initialiser l'architecture du modèle
|
| 23 |
+
|
| 24 |
+
# Charger uniquement les poids du modèle à partir du checkpoint
|
| 25 |
+
model.load_state_dict(checkpoint['model'])
|
| 26 |
+
|
| 27 |
+
model.eval() # Met le modèle en mode évaluation
|
| 28 |
return model
|
| 29 |
|
| 30 |
+
# Prétraitement de l'image
|
| 31 |
def preprocess_frame(frame):
|
| 32 |
+
# Redimensionner et normaliser
|
| 33 |
+
frame = cv2.resize(frame, (224, 224)) # Ajustez la taille si nécessaire
|
| 34 |
+
frame = frame / 255.0 # Normaliser
|
| 35 |
+
input_tensor = torch.from_numpy(frame.astype(np.float32)).permute(2, 0, 1) # Convertir en format Tensor
|
| 36 |
+
return input_tensor.unsqueeze(0) # Ajouter une dimension de lot
|
| 37 |
|
| 38 |
+
# Traitement de la vidéo
|
| 39 |
def process_video(model, video_path):
|
| 40 |
cap = cv2.VideoCapture(video_path)
|
| 41 |
fourcc = cv2.VideoWriter_fourcc(*'mp4v')
|
|
|
|
| 47 |
if not ret:
|
| 48 |
break
|
| 49 |
|
| 50 |
+
# Prétraiter le cadre
|
| 51 |
input_tensor = preprocess_frame(frame)
|
| 52 |
|
| 53 |
+
# Faire des prédictions
|
| 54 |
with torch.no_grad():
|
| 55 |
predictions = model(input_tensor)
|
| 56 |
|
| 57 |
+
# Traiter les prédictions et convertir en image
|
| 58 |
output_frame = (predictions.squeeze().permute(1, 2, 0).numpy() * 255).astype(np.uint8)
|
| 59 |
|
| 60 |
+
# Écrire le cadre traité dans la sortie
|
| 61 |
out.write(output_frame)
|
| 62 |
|
| 63 |
cap.release()
|
| 64 |
out.release()
|
| 65 |
return output_path
|
| 66 |
|
| 67 |
+
# Interface Gradio
|
| 68 |
def colorize_video(video):
|
| 69 |
model = load_model(MODEL_PATH)
|
| 70 |
+
output_video_path = process_video(model, video.name) # Utiliser le nom pour lire la vidéo
|
| 71 |
return output_video_path
|
| 72 |
|
| 73 |
+
# Configuration de l'interface Gradio
|
| 74 |
iface = gr.Interface(
|
| 75 |
fn=colorize_video,
|
| 76 |
+
inputs=gr.Video(label="Téléchargez une vidéo"),
|
| 77 |
+
outputs=gr.Video(label="Vidéo colorisée"),
|
| 78 |
+
title="Colorisation de Vidéos",
|
| 79 |
+
description="Chargez une vidéo en noir et blanc et utilisez le modèle de colorisation pour obtenir une vidéo colorisée."
|
| 80 |
)
|
| 81 |
|
| 82 |
if __name__ == '__main__':
|