Update app.py
Browse files
app.py
CHANGED
@@ -4,12 +4,20 @@ import numpy as np
|
|
4 |
from PIL import Image
|
5 |
import io
|
6 |
import base64
|
|
|
|
|
7 |
|
8 |
app = FastAPI()
|
9 |
|
|
|
|
|
|
|
|
|
|
|
10 |
@app.post("/detect/")
|
11 |
async def detect_face(file: UploadFile = File(...)):
|
12 |
try:
|
|
|
13 |
image_bytes = await file.read()
|
14 |
image = Image.open(io.BytesIO(image_bytes))
|
15 |
img_np = np.array(image)
|
@@ -17,6 +25,7 @@ async def detect_face(file: UploadFile = File(...)):
|
|
17 |
if img_np.shape[2] == 4:
|
18 |
img_np = cv2.cvtColor(img_np, cv2.COLOR_BGRA2BGR)
|
19 |
|
|
|
20 |
face_cascade = cv2.CascadeClassifier(cv2.data.haarcascades + 'haarcascade_frontalface_default.xml')
|
21 |
gray = cv2.cvtColor(img_np, cv2.COLOR_BGR2GRAY)
|
22 |
faces = face_cascade.detectMultiScale(gray, scaleFactor=1.1, minNeighbors=5, minSize=(30, 30))
|
@@ -24,19 +33,38 @@ async def detect_face(file: UploadFile = File(...)):
|
|
24 |
if len(faces) == 0:
|
25 |
raise HTTPException(status_code=404, detail="No se detectaron rostros en la imagen.")
|
26 |
|
|
|
27 |
for (x, y, w, h) in faces:
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
28 |
cv2.rectangle(img_np, (x, y), (x+w, y+h), (255, 0, 0), 2)
|
|
|
|
|
|
|
|
|
29 |
|
|
|
30 |
result_image = Image.fromarray(cv2.cvtColor(img_np, cv2.COLOR_BGR2RGB))
|
31 |
img_byte_arr = io.BytesIO()
|
32 |
result_image.save(img_byte_arr, format='JPEG')
|
33 |
img_byte_arr = img_byte_arr.getvalue()
|
34 |
|
|
|
35 |
return {
|
36 |
"message": "Rostros detectados",
|
37 |
-
"
|
38 |
"imagen_base64": base64.b64encode(img_byte_arr).decode('utf-8')
|
39 |
}
|
40 |
|
41 |
except Exception as e:
|
42 |
raise HTTPException(status_code=500, detail=str(e))
|
|
|
|
4 |
from PIL import Image
|
5 |
import io
|
6 |
import base64
|
7 |
+
from transformers import AutoModelForImageClassification, AutoFeatureExtractor
|
8 |
+
import torch
|
9 |
|
10 |
app = FastAPI()
|
11 |
|
12 |
+
# Cargar el modelo preentrenado para clasificaci贸n de g茅nero
|
13 |
+
model_name = "nateraw/bert-imagenet" # Cambiar a un modelo adecuado si se encuentra uno m谩s espec铆fico
|
14 |
+
model = AutoModelForImageClassification.from_pretrained(model_name)
|
15 |
+
feature_extractor = AutoFeatureExtractor.from_pretrained(model_name)
|
16 |
+
|
17 |
@app.post("/detect/")
|
18 |
async def detect_face(file: UploadFile = File(...)):
|
19 |
try:
|
20 |
+
# Leer y preparar la imagen
|
21 |
image_bytes = await file.read()
|
22 |
image = Image.open(io.BytesIO(image_bytes))
|
23 |
img_np = np.array(image)
|
|
|
25 |
if img_np.shape[2] == 4:
|
26 |
img_np = cv2.cvtColor(img_np, cv2.COLOR_BGRA2BGR)
|
27 |
|
28 |
+
# Detecci贸n de rostros con OpenCV
|
29 |
face_cascade = cv2.CascadeClassifier(cv2.data.haarcascades + 'haarcascade_frontalface_default.xml')
|
30 |
gray = cv2.cvtColor(img_np, cv2.COLOR_BGR2GRAY)
|
31 |
faces = face_cascade.detectMultiScale(gray, scaleFactor=1.1, minNeighbors=5, minSize=(30, 30))
|
|
|
33 |
if len(faces) == 0:
|
34 |
raise HTTPException(status_code=404, detail="No se detectaron rostros en la imagen.")
|
35 |
|
36 |
+
result_data = []
|
37 |
for (x, y, w, h) in faces:
|
38 |
+
# Extraer cada rostro
|
39 |
+
face_img = img_np[y:y+h, x:x+w]
|
40 |
+
face_img_pil = Image.fromarray(cv2.cvtColor(face_img, cv2.COLOR_BGR2RGB)).resize((224, 224))
|
41 |
+
|
42 |
+
# Clasificar el rostro
|
43 |
+
inputs = feature_extractor(images=face_img_pil, return_tensors="pt")
|
44 |
+
outputs = model(**inputs)
|
45 |
+
predicted_class = outputs.logits.argmax(dim=-1).item()
|
46 |
+
label = model.config.id2label[predicted_class]
|
47 |
+
|
48 |
+
# Dibujar rect谩ngulo y agregar datos
|
49 |
cv2.rectangle(img_np, (x, y), (x+w, y+h), (255, 0, 0), 2)
|
50 |
+
result_data.append({
|
51 |
+
"coordenadas": [int(x), int(y), int(w), int(h)],
|
52 |
+
"sexo": label
|
53 |
+
})
|
54 |
|
55 |
+
# Preparar imagen resultante
|
56 |
result_image = Image.fromarray(cv2.cvtColor(img_np, cv2.COLOR_BGR2RGB))
|
57 |
img_byte_arr = io.BytesIO()
|
58 |
result_image.save(img_byte_arr, format='JPEG')
|
59 |
img_byte_arr = img_byte_arr.getvalue()
|
60 |
|
61 |
+
# Respuesta
|
62 |
return {
|
63 |
"message": "Rostros detectados",
|
64 |
+
"resultados": result_data,
|
65 |
"imagen_base64": base64.b64encode(img_byte_arr).decode('utf-8')
|
66 |
}
|
67 |
|
68 |
except Exception as e:
|
69 |
raise HTTPException(status_code=500, detail=str(e))
|
70 |
+
|