Update app.py

app.py

@@ -1,15 +1,17 @@
 import streamlit as st
-from transformers import DetrImageProcessor, DetrForObjectDetection
+from transformers import DetrImageProcessor, DetrForObjectDetection, pipeline
 import torch
 from PIL import Image
-import requests
 
-# Load the DETR model and processor
+# Load the DETR model for object detection
 processor = DetrImageProcessor.from_pretrained("facebook/detr-resnet-50", revision="no_timm")
-model = DetrForObjectDetection.from_pretrained("facebook/detr-resnet-50", revision="no_timm")
+detr_model = DetrForObjectDetection.from_pretrained("facebook/detr-resnet-50", revision="no_timm")
 
-st.title("DETR Object Detection with ResNet-50")
-st.write("Upload an image and let the DETR model detect objects in it.")
+# Load an NLP model for summarization (T5-small used as an example)
+summarizer = pipeline("summarization", model="t5-small")
+
+st.title("DETR Object Detection with NLP Summary")
+st.write("Upload an image to detect objects and get a summary of what is detected.")
 
 # File uploader in Streamlit
 uploaded_file = st.file_uploader("Choose an image...", type="jpg")
@@ -21,14 +23,28 @@ if uploaded_file is not None:
 
     # Process the image and perform object detection
     inputs = processor(images=image, return_tensors="pt")
-    outputs = model(**inputs)
+    outputs = detr_model(**inputs)
 
-    # Post-process the results to get bounding boxes and labels with confidence > 0.9
+    # Post-process the results to get bounding boxes and labels with a lower confidence threshold
     target_sizes = torch.tensor([image.size[::-1]])
-    results = processor.post_process_object_detection(outputs, target_sizes=target_sizes, threshold=0.9)[0]
+    results = processor.post_process_object_detection(outputs, target_sizes=target_sizes, threshold=0.5)[0]
 
-    # Display results
+    # Generate descriptions for detected objects
+    descriptions = []
     st.write("Detected objects:")
     for score, label, box in zip(results["scores"], results["labels"], results["boxes"]):
         box = [round(i, 2) for i in box.tolist()]
-        st.write(f"{model.config.id2label[label.item()]}: {round(score.item(), 3)} at location {box}")
+        label_text = detr_model.config.id2label[label.item()]
+        description = f"Detected {label_text} with confidence {round(score.item(), 2)} at location {box}."
+        descriptions.append(description)
+        st.write(description)  # Display each detected object
+
+    # Combine descriptions into a single text input for the summarizer
+    description_text = " ".join(descriptions)
+
+    # Generate a summary using the NLP model
+    summary = summarizer(description_text, max_length=50, min_length=10, do_sample=False)[0]['summary_text']
+    
+    # Display the summary
+    st.subheader("Summary")
+    st.write(summary)