updated app

app.py

@@ -5,7 +5,7 @@ from PIL import Image
 from io import BytesIO
 import base64
 
-#---------------------------------------------------------------------------------------
+# ---------------------------------------------------------------------------------------
 # The session_state function allows us to initialize and save variables across for across
 # session states. This is a valuable feature that enables us to take different actions
 # depending on the state of selected variables in the code. If this is not done then
@@ -16,18 +16,19 @@ import base64
 # previous value (ss.file_uploaded_id) and if they are the same then we know we
 # don't need to call the face detection model again. We just simply need to process
 # the previous set of detections.
-#---------------------------------------------------------------------------------------
+# ---------------------------------------------------------------------------------------
 
 # Create application title and file uploader widget.
 st.title("OpenCV Deep Learning based Face Detection")
-img_file_buffer = st.file_uploader("Choose a file", type=['jpg', 'jpeg', 'png'])
+img_file_buffer = st.file_uploader("Choose a file", type=["jpg", "jpeg", "png"])
 
 # Initialize session state variables
-if 'file_uploaded_name' not in st.session_state:
+if "file_uploaded_name" not in st.session_state:
     st.session_state.file_uploaded_name = None
-if 'detections' not in st.session_state:
+if "detections" not in st.session_state:
     st.session_state.detections = None
 
+
 # Function for detecting faces in an image.
 def detectFaceOpenCVDnn(net, frame):
     # Create a blob from the image and apply some pre-processing.
@@ -38,6 +39,7 @@ def detectFaceOpenCVDnn(net, frame):
     detections = net.forward()
     return detections
 
+
 # Function for annotating the image with bounding boxes for each detected face.
 def process_detections(frame, detections, conf_threshold=0.5):
     bboxes = []
@@ -57,6 +59,7 @@ def process_detections(frame, detections, conf_threshold=0.5):
             cv2.rectangle(frame, (x1, y1), (x2, y2), (0, 255, 0), bb_line_thickness, cv2.LINE_8)
     return frame, bboxes
 
+
 # Function to load the DNN model.
 @st.cache_resource()
 def load_model():
@@ -65,13 +68,13 @@ def load_model():
     net = cv2.dnn.readNetFromCaffe(configFile, modelFile)
     return net
 
+
 # Function to generate a download link for output file.
-def get_image_download_link(img, filename, text):
+def get_image_download_link(img, label, filename):
     buffered = BytesIO()
     img.save(buffered, format="JPEG")
-    img_str = base64.b64encode(buffered.getvalue()).decode()
-    href = f'<a href="data:file/txt;base64,{img_str}" download="{filename}">{text}</a>'
-    return href
+    st.download_button(label, data=buffered.getvalue(), file_name=filename, mime="image/png")
+
 
 net = load_model()
 
@@ -85,11 +88,11 @@ if img_file_buffer is not None:
     # Create placeholders to display input and output images.
     placeholders = st.columns(2)
     # Display Input image in the first placeholder.
-    placeholders[0].image(image, channels='BGR')
+    placeholders[0].image(image, channels="BGR")
     placeholders[0].text("Input Image")
 
     # Create a Slider and get the threshold from the slider.
-    conf_threshold = st.slider("SET Confidence Threshold", min_value=0.0, max_value=1.0, step=.01, value=0.5)
+    conf_threshold = st.slider("SET Confidence Threshold", min_value=0.0, max_value=1.0, step=0.01, value=0.5)
 
     # Check if the loaded image is "new", if so call the face detection model function.
     if file_name != st.session_state.file_uploaded_name:
@@ -105,10 +108,10 @@ if img_file_buffer is not None:
     out_image, _ = process_detections(image, st.session_state.detections, conf_threshold=conf_threshold)
 
     # Display Detected faces.
-    placeholders[1].image(out_image, channels='BGR')
+    placeholders[1].image(out_image, channels="BGR")
     placeholders[1].text("Output Image")
 
     # Convert OpenCV image to PIL.
     out_image = Image.fromarray(out_image[:, :, ::-1])
     # Create a link for downloading the output file.
-    st.markdown(get_image_download_link(out_image, "face_output.jpg", 'Download Output Image'), unsafe_allow_html=True)
+    get_image_download_link(out_image, "Download Output Image", "output.jpg")

requirements.txt

@@ -1,4 +1,4 @@
 numpy
-streamlit
+streamlit==1.44.1
 opencv-python-headless
 pillow