def predict_label():
_data = Data()
labels = _data.labels
if request.method == 'GET':
data = _data.test_data
image_data = get_image_data(data)
prediction = predict_image(image_data)
argmax = int(np.argmax(np.array(prediction)[0]))
return jsonify({labels[argmax]: prediction[0][argmax]})
elif request.method == 'POST':
input_data = request.get_json()
raw_data = input_data["image_data"]
decoded = base64.b64decode(str(raw_data))
io_bytes = io.BytesIO(decoded)
data = Image.open(io_bytes)
image_data = get_image_data(data)
prediction = predict_image(image_data)
argmax = int(np.argmax(np.array(prediction)[0]))
job_id = data['job_id'] if 'job_id' in input_data.keys(
) else get_job_id()
return jsonify({
labels[argmax]: prediction[0][argmax],
'job_id': job_id
})
async def _predict_async_post(
data: Data,
job_id: str,
background_tasks: BackgroundTasks = BackgroundTasks()
) -> Dict[str, List[float]]:
image = base64.b64decode(str(data.image_data))
io_bytes = io.BytesIO(image)
data.image_data = Image.open(io_bytes)
store_data_job._save_data_job(data, job_id, background_tasks, True)
return {'job_id': job_id}
async def _predict_label(
data: Data,
job_id: str,
background_tasks: BackgroundTasks = BackgroundTasks()
) -> Dict[str, List[float]]:
image = base64.b64decode(str(data.image_data))
io_bytes = io.BytesIO(image)
data.image_data = Image.open(io_bytes)
label_proba = await __async_predict_label(data)
store_data_job._save_data_job(data, job_id, background_tasks, False)
return {'prediction': label_proba, 'job_id': job_id}
def predict():
_data = Data()
if request.method == 'GET':
data = _data.test_data
image_data = get_image_data(data)
prediction = predict_image(image_data)
return jsonify({'prediction': prediction})
elif request.method == 'POST':
input_data = request.get_json()
raw_data = input_data["image_data"]
decoded = base64.b64decode(str(raw_data))
io_bytes = io.BytesIO(decoded)
data = Image.open(io_bytes)
image_data = get_image_data(data)
prediction = predict_image(image_data)
job_id = data['job_id'] if 'job_id' in input_data.keys(
) else get_job_id()
return jsonify({'prediction': prediction, 'job_id': job_id})
def predict():
_data = Data()
if request.method == "GET":
data = _data.test_data
image_data = get_image_data(data)
prediction = predict_image(image_data)
return jsonify({"prediction": prediction})
elif request.method == "POST":
input_data = request.get_json()
raw_data = input_data["image_data"]
decoded = base64.b64decode(str(raw_data))
io_bytes = io.BytesIO(decoded)
data = Image.open(io_bytes)
image_data = get_image_data(data)
prediction = predict_image(image_data)
job_id = data["job_id"] if "job_id" in input_data.keys(
) else get_job_id()
return jsonify({"prediction": prediction, "job_id": job_id})
async def _test_label(data: Data = Data()) -> Dict[str, Dict[str, float]]:
data.image_data = data.test_data
label_proba = __predict_label(data)
return {'prediction': label_proba}
async def _test(data: Data = Data()) -> Dict[str, int]:
data.image_data = data.test_data
__predict(data)
return {'prediction': data.prediction}
async def __async_predict(data: Data):
image_data = get_image_data(data.image_data)
output_np = await active_predictor.async_predict(image_data)
reshaped_output_nps = DataConverter.reshape_output(output_np)
data.prediction = reshaped_output_nps.tolist()
logger.info({'job_id': data.job_id, 'prediction': data.prediction})
def __predict(data: Data):
input_np = DataConverter.convert_input_data_to_np(data.input_data)
output_np = active_predictor.predict(input_np)
reshaped_output_nps = DataConverter.reshape_output(output_np)
data.prediction = reshaped_output_nps.tolist()
logger.info({'job_id': data.job_id, 'prediction': data.prediction})
def labels():
_data = Data()
labels = _data.labels
return jsonify({'labels': labels})
def labels():
_data = Data()
labels = _data.labels
return jsonify({"labels": labels})
def _test_label(data: Data = Data()) -> Dict[str, Dict[str, float]]:
data.input_data = data.test_data
label_proba = __predict_label(data)
return {"prediction": label_proba}
def _test(data: Data = Data()) -> Dict[str, int]:
data.input_data = data.test_data
__predict(data)
return {"prediction": data.prediction}
async def __async_predict(data: Data):
input_np = DataConverter.convert_input_data_to_np(data.input_data)
output_np = await active_predictor.async_predict(input_np)
reshaped_output_nps = DataConverter.reshape_output(output_np)
data.prediction = reshaped_output_nps.tolist()
logger.info({"job_id": data.job_id, "prediction": data.prediction})