first commit

app.py

@@ -1,8 +1,92 @@
-import time
+import os
+import torch
 import streamlit as st
+from transformers import pipeline, AutoImageProcessor
 from PIL import Image
+from utils import download_model_from_s3
+
+model_path = {
+    "TinyBert Sentiment Analysis": "ml-models/tinybert-sentiment-analysis/",
+    "TinyBert Disaster Classification": "ml-models/tinybert-disaster-tweet/",
+    "VIT Pose Classification": "ml-models/vit-human-pose-classification/"
+}
+
 
 st.title("Machine Learning Model Deployment")
-x = st.slider('Select a value')
-st.write(x, 'squared is', x * x)
+
+model_choice = st.selectbox(
+    "Select Model:",[
+        "TinyBert Sentiment Analysis",
+        "TinyBert Disaster Classification",
+        "VIT Pose Classification"   
+    ]
+)
+
+local_path = model_choice.lower().replace(" ", "-")
+s3_prefix = model_paths[model_choice]
+device = torch.device("cuda") if torch.cuda.is_available() else torch.device("cpu")
+
+if "downloaded_models" not in st.session_state:
+    st.session_state.downloaded_models = set()
+
+
+if model_choice not in st.session_state.downloaded_models:
+    if st.button(f"Download {model_choice}"):
+        with st.spinner(f"Downloading {model_choice}... Please wait!"):
+            download_model_from_s3(local_path, s3_prefix)
+        st.session_state.downloaded_models.add(model_choice)
+        st.toast(f"✅ {model_choice} Succesfuly Download!", icon="🎉")
+
+# **1. Sentiment Analysis Model**
+if model_choice == "TinyBERT Sentiment Analysis":
+    text = st.text_area("Enter Text:", "Your review...")
+    predict = st.button("Predict Sentiment")
+
+    classifier = pipeline("text-classification", model=local_path, device=device)
+
+    if predict:
+        with st.spinner("Predicting..."):
+            output = classifier(text)
+            st.write(output)
+
+# **2. Disaster Classification**
+if model_choice == "TinyBert Disaster Classification":
+    text = st.text_area("Enter Text:", "Your Tweet...")
+    predict = st.button("Predict Sentiment")
+
+    classifier = pipeline("text-classification", model=local_path, device=device)
+
+    if predict:
+        with st.spinner("Predicting..."):
+            output = classifier(text)
+            st.write(output)
+
+# **3. Image Classification**
+if model_choice == "VIT Pose Classification":
+    uploaded_file = st.file_uploader("Upload Image", type=["jpg", "png", "jpeg"])
+    predict = st.button("Predict Image")
+
+    if uploaded_file is not None:
+        image = Image.open(uploaded_file)
+        st.image(image, caption="Your Image", use_column_width=True)
+
+        image_processor = AutoImageProcessor.from_pretrained(local_directory, use_fast=True)
+        pipe = pipeline('image-classification', model=local_path, image_processor=image_processor, device=device)
+
+        if predict:
+            with st.spinner("Predicting..."):
+                output = pipe(image)
+                st.write(output)
+
+
+
+
+user_input = st.text_area("Enter text here...")
+predict = st.button("Predict")
+if predict:
+    with st.spinner("Downloading Model...")
+    
+
+
+
 

requirements.txt

@@ -1 +1,4 @@
 streamlit
+torch
+boto3
+transformers

utils.py

@@ -0,0 +1,43 @@
+import os
+import boto3
+from path import Path
+
+# credentials aws
+aws_access_key = os.getenv("AWS_ACCESS_KEY_ID")
+aws_secret_key = os.getenv("AWS_SECRET_ACCESS_KEY")
+bucket_name = os.getenv("BUCKET_NAME")
+
+
+s3 = boto3.client(
+    "s3",
+    aws_access_key_id=aws_access_key,
+    aws_secret_access_key=aws_secret_key
+)
+
+
+def download_model_from_s3(local_path: Path, s3_prefix: str):
+    if os.path.exists(local_path) and os.listdir(local_path):
+        st.toast(f"✅ Model {local_path} Available!", icon="🎉")
+        return
+
+    os.makedirs(local_path, exist_ok=True)
+    paginator = s3.get_paginator("list_objects_v2")
+
+    for result in paginator.paginate(Bucket=bucket_name, Prefix=s3_prefix):
+        if "Contents" in result:
+            for key in result["Contents"]:
+                s3_key = key["Key"]
+                local_file = os.path.join(local_path, os.path.relpath(s3_key, s3_prefix))
+               
+                os.makedirs(os.path.dirname(local_file), exist_ok=True)
+                s3.download_file(bucket_name, s3_key, local_file)
+
+
+
+
+
+
+
+
+
+