Spaces:
Runtime error
Runtime error
Lambang
commited on
Commit
•
bb9d3db
1
Parent(s):
1c26982
daaaa
Browse files
main.py
CHANGED
@@ -1,9 +1,7 @@
|
|
1 |
-
from fastapi import FastAPI
|
2 |
-
import pickle
|
3 |
-
import uvicorn
|
4 |
-
import pandas as pd
|
5 |
import shutil
|
6 |
import cv2
|
|
|
|
|
7 |
import tensorflow as tf
|
8 |
import os
|
9 |
from flask import Flask, jsonify, request, flash, redirect, url_for
|
@@ -11,12 +9,9 @@ from pyngrok import ngrok
|
|
11 |
from fastapi import FastAPI, HTTPException, File, UploadFile, Request
|
12 |
from fastapi.staticfiles import StaticFiles
|
13 |
from fastapi.responses import JSONResponse
|
14 |
-
|
15 |
-
|
16 |
-
from
|
17 |
-
from get_load_data import GetLoadData
|
18 |
-
from data_preprocess import DataProcessing
|
19 |
-
from train_pred import TrainPred
|
20 |
|
21 |
app = FastAPI()
|
22 |
public_url = "https://lambang0902-test-space.hf.space"
|
@@ -26,37 +21,38 @@ app.mount("/static", StaticFiles(directory="static"), name="static")
|
|
26 |
filepath = ""
|
27 |
list_class = ['Diamond','Oblong','Oval','Round','Square','Triangle']
|
28 |
list_folder = ['Training', 'Testing']
|
|
|
|
|
29 |
face_crop_img = True
|
30 |
face_landmark_img = True
|
31 |
landmark_extraction_img = True
|
32 |
-
|
33 |
|
34 |
-
|
35 |
# Tempat deklarasi model dan sejenisnya
|
36 |
selected_model = tf.keras.models.load_model(f'models/fc_model_1.h5', compile=False)
|
37 |
-
|
38 |
-
|
39 |
-
|
40 |
-
|
41 |
-
#
|
42 |
|
43 |
|
44 |
-
#
|
45 |
# Tempat setting server
|
46 |
UPLOAD_FOLDER = './upload'
|
47 |
UPLOAD_MODEL = './models'
|
48 |
ALLOWED_EXTENSIONS = {'png', 'jpg', 'jpeg','zip','h5'}
|
49 |
-
#
|
50 |
-
#
|
51 |
-
#
|
52 |
-
|
53 |
-
async def root():
|
54 |
-
# Dapatkan URL publik dari ngrok
|
55 |
-
ngrok_url = "Tidak Ada URL Publik (ngrok belum selesai memulai)"
|
56 |
-
|
57 |
-
return {"message": "Hello, World!", "ngrok_url": ngrok_url}
|
58 |
-
|
59 |
|
|
|
|
|
|
|
|
|
|
|
60 |
#-----------------------------------------------------
|
61 |
|
62 |
data_processor = DataProcessing()
|
@@ -69,7 +65,6 @@ def preprocessing(filepath):
|
|
69 |
shutil.rmtree(folder_path)
|
70 |
os.mkdir(folder_path)
|
71 |
|
72 |
-
|
73 |
data_processor.detect_landmark(data_processor.face_cropping_pred(filepath))
|
74 |
# data_processor.enhance_contrast_histeq(data_processor.face_cropping_pred(filepath))
|
75 |
|
@@ -81,14 +76,40 @@ def preprocessing(filepath):
|
|
81 |
os.rename(os.path.join(folder_path, file_name), os.path.join(folder_path, new_file_name))
|
82 |
index += 1
|
83 |
|
84 |
-
|
85 |
|
86 |
-
|
87 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
88 |
|
89 |
-
|
90 |
-
|
91 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
92 |
|
93 |
@app.post('/upload/file',tags=["Predicting"])
|
94 |
async def upload_file(picture: UploadFile):
|
@@ -99,14 +120,12 @@ async def upload_file(picture: UploadFile):
|
|
99 |
raise HTTPException(status_code=400, detail='Invalid file extension')
|
100 |
|
101 |
os.makedirs(UPLOAD_FOLDER, exist_ok=True)
|
102 |
-
file_path = os.path.join(UPLOAD_FOLDER, (picture.filename))
|
103 |
-
|
104 |
-
|
105 |
with open(file_path, 'wb') as f:
|
106 |
f.write(picture.file.read())
|
107 |
try:
|
108 |
-
|
109 |
-
|
110 |
except Exception as e:
|
111 |
os.remove(file_path)
|
112 |
raise HTTPException(status_code=500, detail=f'Error processing image: {str(e)}')
|
@@ -118,14 +137,185 @@ def get_images():
|
|
118 |
folder_path = "./static/temporary"
|
119 |
files = [f for f in os.listdir(folder_path) if os.path.isfile(os.path.join(folder_path, f))]
|
120 |
urls = []
|
|
|
|
|
121 |
for i in range(0, 3):
|
122 |
url = f'{public_url}/static/temporary/{files[i]}'
|
123 |
urls.append(url)
|
124 |
-
|
125 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
126 |
|
127 |
if __name__ == '__main__':
|
128 |
import uvicorn
|
129 |
public_url = ngrok.connect(8080).public_url
|
130 |
print(f' * Running on {public_url}')
|
131 |
-
uvicorn.run(app, host="0.0.0.0", port=8080)
|
|
|
|
|
|
|
|
|
|
|
|
1 |
import shutil
|
2 |
import cv2
|
3 |
+
import mediapipe as mp
|
4 |
+
from werkzeug.utils import secure_filename
|
5 |
import tensorflow as tf
|
6 |
import os
|
7 |
from flask import Flask, jsonify, request, flash, redirect, url_for
|
|
|
9 |
from fastapi import FastAPI, HTTPException, File, UploadFile, Request
|
10 |
from fastapi.staticfiles import StaticFiles
|
11 |
from fastapi.responses import JSONResponse
|
12 |
+
from pydantic import BaseModel
|
13 |
+
import subprocess
|
14 |
+
from hairstyle_recommendation import HairstyleRecommendation
|
|
|
|
|
|
|
15 |
|
16 |
app = FastAPI()
|
17 |
public_url = "https://lambang0902-test-space.hf.space"
|
|
|
21 |
filepath = ""
|
22 |
list_class = ['Diamond','Oblong','Oval','Round','Square','Triangle']
|
23 |
list_folder = ['Training', 'Testing']
|
24 |
+
recommendation = HairstyleRecommendation()
|
25 |
+
|
26 |
face_crop_img = True
|
27 |
face_landmark_img = True
|
28 |
landmark_extraction_img = True
|
29 |
+
#-----------------------------------------------------
|
30 |
|
31 |
+
#-----------------------------------------------------
|
32 |
# Tempat deklarasi model dan sejenisnya
|
33 |
selected_model = tf.keras.models.load_model(f'models/fc_model_1.h5', compile=False)
|
34 |
+
face_cascade = cv2.CascadeClassifier('haarcascade_frontalface_alt2.xml')
|
35 |
+
mp_drawing = mp.solutions.drawing_utils
|
36 |
+
mp_face_mesh = mp.solutions.face_mesh
|
37 |
+
drawing_spec = mp_drawing.DrawingSpec(thickness=1, circle_radius=1)
|
38 |
+
# #-----------------------------------------------------
|
39 |
|
40 |
|
41 |
+
# #-----------------------------------------------------
|
42 |
# Tempat setting server
|
43 |
UPLOAD_FOLDER = './upload'
|
44 |
UPLOAD_MODEL = './models'
|
45 |
ALLOWED_EXTENSIONS = {'png', 'jpg', 'jpeg','zip','h5'}
|
46 |
+
# app.config['UPLOAD_FOLDER'] = UPLOAD_FOLDER
|
47 |
+
# app.config['UPLOAD_MODEL'] = UPLOAD_MODEL
|
48 |
+
# app.config['MAX_CONTENT_LENGTH'] = 500 * 1024 * 1024 # 500 MB
|
49 |
+
# #-----------------------------------------------------
|
|
|
|
|
|
|
|
|
|
|
|
|
50 |
|
51 |
+
#
|
52 |
+
from file_processing import FileProcess
|
53 |
+
from get_load_data import GetLoadData
|
54 |
+
from data_preprocess import DataProcessing
|
55 |
+
from train_pred import TrainPred
|
56 |
#-----------------------------------------------------
|
57 |
|
58 |
data_processor = DataProcessing()
|
|
|
65 |
shutil.rmtree(folder_path)
|
66 |
os.mkdir(folder_path)
|
67 |
|
|
|
68 |
data_processor.detect_landmark(data_processor.face_cropping_pred(filepath))
|
69 |
# data_processor.enhance_contrast_histeq(data_processor.face_cropping_pred(filepath))
|
70 |
|
|
|
76 |
os.rename(os.path.join(folder_path, file_name), os.path.join(folder_path, new_file_name))
|
77 |
index += 1
|
78 |
|
79 |
+
print("Tungu sampai selesaiii")
|
80 |
|
81 |
+
train_datagen = tf.keras.preprocessing.image.ImageDataGenerator(rescale=1 / 255.)
|
82 |
+
test_datagen = tf.keras.preprocessing.image.ImageDataGenerator(rescale=1 / 255.)
|
83 |
+
|
84 |
+
#-----------------------------------------------------
|
85 |
+
# Fungsi untuk menjalankan ngrok
|
86 |
+
def run_ngrok():
|
87 |
+
try:
|
88 |
+
# Jalankan ngrok dan simpan prosesnya
|
89 |
+
ngrok_process = subprocess.Popen(['ngrok', 'http', '8000'])
|
90 |
+
return ngrok_process
|
91 |
+
except Exception as e:
|
92 |
+
print(f"Error running ngrok: {e}")
|
93 |
|
94 |
+
|
95 |
+
@app.get("/")
|
96 |
+
async def root():
|
97 |
+
# Dapatkan URL publik dari ngrok
|
98 |
+
ngrok_url = "Tidak Ada URL Publik (ngrok belum selesai memulai)"
|
99 |
+
try:
|
100 |
+
ngrok_url = subprocess.check_output(['ngrok', 'http', '8000']).decode('utf-8').strip()
|
101 |
+
except Exception as e:
|
102 |
+
print(f"Error getting ngrok URL: {e}")
|
103 |
+
|
104 |
+
return {"message": "Hello, World!", "ngrok_url": ngrok_url}
|
105 |
+
|
106 |
+
# -------------------------------------------------------------------------
|
107 |
+
# API UNTUK MELAKUKAN PROSES PREDIKSI
|
108 |
+
# -------------------------------------------------------------------------
|
109 |
+
# Use a pipeline as a high-level helper
|
110 |
+
from transformers import pipeline
|
111 |
+
|
112 |
+
pipe = pipeline("image-classification", model="rizvandwiki/gender-classification-2")
|
113 |
|
114 |
@app.post('/upload/file',tags=["Predicting"])
|
115 |
async def upload_file(picture: UploadFile):
|
|
|
120 |
raise HTTPException(status_code=400, detail='Invalid file extension')
|
121 |
|
122 |
os.makedirs(UPLOAD_FOLDER, exist_ok=True)
|
123 |
+
file_path = os.path.join(UPLOAD_FOLDER, secure_filename(picture.filename))
|
|
|
|
|
124 |
with open(file_path, 'wb') as f:
|
125 |
f.write(picture.file.read())
|
126 |
try:
|
127 |
+
preprocessing(cv2.imread(file_path))
|
128 |
+
|
129 |
except Exception as e:
|
130 |
os.remove(file_path)
|
131 |
raise HTTPException(status_code=500, detail=f'Error processing image: {str(e)}')
|
|
|
137 |
folder_path = "./static/temporary"
|
138 |
files = [f for f in os.listdir(folder_path) if os.path.isfile(os.path.join(folder_path, f))]
|
139 |
urls = []
|
140 |
+
|
141 |
+
# Image preprocessed url
|
142 |
for i in range(0, 3):
|
143 |
url = f'{public_url}/static/temporary/{files[i]}'
|
144 |
urls.append(url)
|
145 |
+
|
146 |
+
# Face shape classification
|
147 |
+
bentuk, persentase = data_train_pred.prediction(selected_model)
|
148 |
+
|
149 |
+
# Gender classification
|
150 |
+
gender_classify = pipe('./static/result_upload0.jpg')
|
151 |
+
output_gender = max(gender_classify, key=lambda x: x['score'])['label']
|
152 |
+
|
153 |
+
|
154 |
+
# Hairstyle recommendation
|
155 |
+
recommended_styles, style_images = recommendation.get_recommendation(output_gender, bentuk[0])
|
156 |
+
hairstyleImage = [f'{public_url}/static/hairstyle_image/{file}' for file in style_images]
|
157 |
+
|
158 |
+
response = {'urls': urls,
|
159 |
+
'bentuk_wajah': bentuk[0],
|
160 |
+
'persen': persentase,
|
161 |
+
'gender': output_gender,
|
162 |
+
'hair_style': recommended_styles,
|
163 |
+
'hair_image': hairstyleImage}
|
164 |
+
return response
|
165 |
+
|
166 |
+
|
167 |
+
# -------------------------------------------------------------------------
|
168 |
+
# API UNTUK MELAKUKAN PROSES TRAINING
|
169 |
+
# -------------------------------------------------------------------------
|
170 |
+
|
171 |
+
# Model pydantic untuk validasi body
|
172 |
+
class TrainingParams(BaseModel):
|
173 |
+
optimizer: str
|
174 |
+
epoch: int
|
175 |
+
batchSize: int
|
176 |
+
|
177 |
+
@app.post('/upload/dataset', tags=["Training"])
|
178 |
+
async def upload_data(dataset: UploadFile):
|
179 |
+
if dataset.filename == '':
|
180 |
+
raise HTTPException(status_code=400, detail='No file selected for uploading')
|
181 |
+
|
182 |
+
# Buat path lengkap untuk menyimpan file
|
183 |
+
file_path = os.path.join(UPLOAD_FOLDER, dataset.filename)
|
184 |
+
|
185 |
+
# Simpan file ke folder yang ditentukan
|
186 |
+
with open(file_path, "wb") as file_object:
|
187 |
+
file_object.write(dataset.file.read())
|
188 |
+
|
189 |
+
# Panggil fungsi untuk mengekstrak file jika perlu
|
190 |
+
FileProcess.extract_zip(file_path)
|
191 |
+
|
192 |
+
return {'message': 'File successfully uploaded'}
|
193 |
+
|
194 |
+
@app.post('/set_params', tags=["Training"])
|
195 |
+
async def set_params(request: Request, params: TrainingParams):
|
196 |
+
global optimizer, epoch, batch_size
|
197 |
+
|
198 |
+
optimizer = params.optimizer
|
199 |
+
epoch = params.epoch
|
200 |
+
batch_size = params.batchSize
|
201 |
+
|
202 |
+
response = {'message': 'Set parameter sukses'}
|
203 |
+
return response
|
204 |
+
|
205 |
+
@app.get('/get_info_data', tags=["Training"])
|
206 |
+
def get_info_prepro():
|
207 |
+
global optimizer, epoch, batch_size
|
208 |
+
training_counts = GetLoadData.get_training_file_counts().json
|
209 |
+
testing_counts = GetLoadData.get_testing_file_counts().json
|
210 |
+
response = {
|
211 |
+
"optimizer": optimizer,
|
212 |
+
"epoch": epoch,
|
213 |
+
"batch_size": batch_size,
|
214 |
+
"training_counts": training_counts,
|
215 |
+
"testing_counts": testing_counts
|
216 |
+
}
|
217 |
+
return response
|
218 |
+
|
219 |
+
@app.get('/get_images_preprocess', tags=["Training"])
|
220 |
+
def get_random_images_crop():
|
221 |
+
images_face_landmark = GetLoadData.get_random_images(tahap="Face Landmark",public_url=public_url)
|
222 |
+
images_face_extraction = GetLoadData.get_random_images(tahap="landmark Extraction", public_url=public_url)
|
223 |
+
|
224 |
+
response = {
|
225 |
+
"face_landmark": images_face_landmark,
|
226 |
+
"landmark_extraction": images_face_extraction
|
227 |
+
}
|
228 |
+
return response
|
229 |
+
|
230 |
+
@app.get('/do_preprocessing', tags=["Training"])
|
231 |
+
async def do_preprocessing():
|
232 |
+
try:
|
233 |
+
data_train_pred.do_pre1(test="")
|
234 |
+
data_train_pred.do_pre2(test="")
|
235 |
+
return {'message': 'Preprocessing sukses'}
|
236 |
+
except Exception as e:
|
237 |
+
# Tangani kesalahan dan kembalikan respons kesalahan
|
238 |
+
error_message = f'Error during preprocessing: {str(e)}'
|
239 |
+
raise HTTPException(status_code=500, detail=error_message)
|
240 |
+
|
241 |
+
@app.get('/do_training', tags=["Training"])
|
242 |
+
def do_training():
|
243 |
+
global epoch
|
244 |
+
folder = ""
|
245 |
+
if (face_landmark_img == True and landmark_extraction_img == True):
|
246 |
+
folder = "Landmark Extraction"
|
247 |
+
elif (face_landmark_img == True and landmark_extraction_img == False):
|
248 |
+
folder = "Face Landmark"
|
249 |
+
# --------------------------------------------------------------
|
250 |
+
train_dataset_path = f"./static/dataset/{folder}/Training/"
|
251 |
+
test_dataset_path = f"./static/dataset/{folder}/Testing/"
|
252 |
+
|
253 |
+
train_image_df, test_image_df = GetLoadData.load_image_dataset(train_dataset_path, test_dataset_path)
|
254 |
+
|
255 |
+
train_gen, test_gen = data_train_pred.data_configuration(train_image_df, test_image_df)
|
256 |
+
model = data_train_pred.model_architecture()
|
257 |
+
|
258 |
+
result = data_train_pred.train_model(model, train_gen, test_gen, epoch)
|
259 |
+
|
260 |
+
# Mengambil nilai akurasi training dan validation dari objek result
|
261 |
+
train_acc = result.history['accuracy'][-1]
|
262 |
+
val_acc = result.history['val_accuracy'][-1]
|
263 |
+
|
264 |
+
# Plot accuracy
|
265 |
+
data_train_pred.plot_accuracy(result=result, epoch=epoch)
|
266 |
+
acc_url = f'{public_url}/static/accuracy_plot.png'
|
267 |
+
|
268 |
+
# Plot loss
|
269 |
+
data_train_pred.plot_loss(result=result, epoch=epoch)
|
270 |
+
loss_url = f'{public_url}/static/loss_plot.png'
|
271 |
+
|
272 |
+
# Confusion Matrix
|
273 |
+
data_train_pred.plot_confusion_matrix(model, test_gen)
|
274 |
+
conf_url = f'{public_url}/static/confusion_matrix.png'
|
275 |
+
|
276 |
+
return jsonify({'train_acc': train_acc, 'val_acc': val_acc, 'plot_acc': acc_url, 'plot_loss':loss_url,'conf':conf_url})
|
277 |
+
|
278 |
+
|
279 |
+
# -------------------------------------------------------------------------
|
280 |
+
# API UNTUK PEMILIHAN MODEL
|
281 |
+
# -------------------------------------------------------------------------
|
282 |
+
|
283 |
+
|
284 |
+
@app.post('/upload/model', tags=["Model"])
|
285 |
+
def upload_model():
|
286 |
+
if 'file' not in request.files:
|
287 |
+
return {'message': 'No file part in the request'}, 400
|
288 |
+
|
289 |
+
file = request.files['file']
|
290 |
+
|
291 |
+
if file.filename == '':
|
292 |
+
return {'message': 'No file selected for uploading'}, 400
|
293 |
+
|
294 |
+
if file and FileProcess.allowed_file(file.filename):
|
295 |
+
filename = secure_filename(file.filename)
|
296 |
+
filepath = os.path.join(app.config['UPLOAD_MODEL'], filename)
|
297 |
+
file.save(filepath)
|
298 |
+
|
299 |
+
return {'message': 'File successfully uploaded'}
|
300 |
+
|
301 |
+
return {'message': 'File failed to uploaded'}
|
302 |
+
|
303 |
+
@app.post('/selected_models')
|
304 |
+
def select_models(index: int):
|
305 |
+
global selected_model
|
306 |
+
try:
|
307 |
+
global selected_model
|
308 |
+
selected_model = tf.keras.models.load_model(f'models/fc_model_{index}.h5')
|
309 |
+
|
310 |
+
# Lakukan sesuatu dengan indeks yang diterima
|
311 |
+
|
312 |
+
return {'message': 'Request berhasil diterima'}
|
313 |
+
except Exception as e:
|
314 |
+
raise HTTPException(status_code=500, detail=f'Error: {str(e)}')
|
315 |
|
316 |
if __name__ == '__main__':
|
317 |
import uvicorn
|
318 |
public_url = ngrok.connect(8080).public_url
|
319 |
print(f' * Running on {public_url}')
|
320 |
+
uvicorn.run(app, host="0.0.0.0", port=8080)
|
321 |
+
# app = FastAPI()
|