def predict():
if model:
try:
input_df = pd.DataFrame(request.json)
predictions = model_utils.predict(input_df, model)
return jsonify(predictions)
except Exception as e:
return jsonify({'error': str(e), 'trace': traceback.format_exc()})
else:
print('You need to train a model before you can make predictions.')
return 'error: no model'
def predict():
print(model)
if model:
try:
input_df = pd.DataFrame(request.json)
predictions = model_utils.predict(input_df, model, model_columns)
return jsonify(predictions)
except Exception as e:
return jsonify({'error': str(e), 'trace': traceback.format_exc()})
else:
return jsonify({'error': 'Please train model before trying to predict'})
def predict():
print("Data is\n", request.json)
input_df=model_utils.transform(request.json)
print("Data transformed\n",input_df)
if model:
print("model exists")
try:
#input_df = pd.DataFrame(request.json)
predictions = model_utils.predict(input_df, model)
#print("Predictions", predictions)
return jsonify(predictions)
except Exception as e:
return jsonify({'error': str(e), 'trace': traceback.format_exc()})
else:
print('You need to train a model before you can make predictions.')
return 'error: no model'
@st.cache
def model_loader(config):
model = load_model(config["paths"]["model_path"])
return model
model = model_loader(config)
st.write("""
# Multi task classifier
## Enter the image url
""")
url = st.text_input("Enter image url")
if url:
current_image = download_image(url)
predictions = predict(model, current_image, config)
col1, col2 = st.beta_columns(2)
col1.write("Original image")
col1.image(current_image, use_column_width=True)
for encoder in predictions.keys():
current_dict = predictions[encoder]
current_df = pd.DataFrame(
index=current_dict.keys(),
data=current_dict.values(),
columns=["prediction"],
).sort_values(by="prediction", ascending=False)
col2.write(f"{encoder} prediction")
col2.dataframe(current_df)