async def predict_scene(image: UploadFile = File(...)):
try:
contents = await image.read()
image = Image.open(io.BytesIO(contents))
resized_image = image.resize((scene_input_width, scene_input_height),
Image.ANTIALIAS)
results = classify_image(scene_interpreter, image=resized_image)
label_id, prob = results[0]
data = {}
data["label"] = scene_labels[label_id]
data["confidence"] = prob
data["success"] = True
return data
except:
e = sys.exc_info()[1]
raise HTTPException(status_code=500, detail=str(e))
async def predict_face(image: UploadFile = File(...)):
try:
contents = await image.read()
image = Image.open(io.BytesIO(contents))
image_width = image.size[0]
image_height = image.size[1]
# Format data and send to interpreter
resized_image = image.resize((face_input_width, face_input_height),
Image.ANTIALIAS)
input_data = np.expand_dims(resized_image, axis=0)
face_interpreter.set_tensor(face_input_details[0]["index"], input_data)
# Process image and get predictions
face_interpreter.invoke()
boxes = face_interpreter.get_tensor(face_output_details[0]["index"])[0]
classes = face_interpreter.get_tensor(
face_output_details[1]["index"])[0]
scores = face_interpreter.get_tensor(
face_output_details[2]["index"])[0]
data = {}
faces = []
for i in range(len(scores)):
if not classes[i] == 0: # Face
continue
single_face = {}
single_face["userid"] = "unknown"
single_face["confidence"] = float(scores[i])
single_face["y_min"] = int(float(boxes[i][0]) * image_height)
single_face["x_min"] = int(float(boxes[i][1]) * image_width)
single_face["y_max"] = int(float(boxes[i][2]) * image_height)
single_face["x_max"] = int(float(boxes[i][3]) * image_width)
if single_face["confidence"] < MIN_CONFIDENCE:
continue
faces.append(single_face)
data["predictions"] = faces
data["success"] = True
return data
except:
e = sys.exc_info()[1]
raise HTTPException(status_code=500, detail=str(e))