import streamlit as st import cv2 import tempfile from ultralytics import YOLO import numpy as np import time alerting_classes = { 0: 'People', 2: 'Car', 7: 'Truck', 24: 'Backpack', 65: 'Suspicious handheld device', 26: 'Handbag', 28: 'Suitcase', } red_tint = np.array([[[0, 0, 255]]], dtype=np.uint8) model1 = YOLO('yolov8n.pt') st.title("Object Detection and Recognition") st.write(""" This web app performs object detection and recognition on a video using YOLOv8. It detects various objects, such as people, cars, trucks, backpacks, suspicious handheld devices, handbags, and suitcases. The processed video is displayed with alerts highlighted, and you can stop the inference at any time. """) video_file = st.file_uploader("Choose a video file", type=["mp4"]) video_placeholder = st.image([]) results = None centered_text = """
Built with ❤️ by Unnati
""" if video_file is not None: # Create temporary file for uploaded video tfile = tempfile.NamedTemporaryFile(delete=False) tfile.write(video_file.read()) # Open video capture using temporary file path # Open video capture using temporary file path cap = cv2.VideoCapture(tfile.name) original_width = int(cap.get(cv2.CAP_PROP_FRAME_WIDTH)) original_height = int(cap.get(cv2.CAP_PROP_FRAME_HEIGHT)) # Set the target width and height based on the conditions target_width = int(original_width * 0.65) if original_width <= 1920 else int(original_width * 0.5) target_height = int(original_height * 0.65) if original_width <= 1920 else int(original_height * 0.5) alert_set = set(alerting_classes.keys()) alert_set.remove(0) # Create red-tinted overlay red_tinted_overlay = np.tile(red_tint, (1, 1, 1)) stop_button = st.button("Stop Inference") # Collect frames in a list frames = [] frame_counter = 0 # Counter to track frame number total_frames = int(cap.get(cv2.CAP_PROP_FRAME_COUNT)) progress_bar_processing = st.progress(0) while cap.isOpened() and not stop_button: success, frame = cap.read() # if the frame is read correctly ret is True if not success: # st.warning("Can't receive frame (stream end?). Exiting ...") break # Resize the frame resized_frame = cv2.resize(frame, (target_width, target_height)) if frame_counter % 4 == 0: # Perform inference on every 4th frame alert_flag = False alert_reason = [] # Perform YOLO object detection results = model1(frame, conf=0.35, verbose=False, classes=list(alerting_classes.keys())) class_ids = results[0].boxes.cls.tolist() class_counts = {cls: class_ids.count(cls) for cls in set(class_ids)} for cls in alert_set: if cls in class_counts and class_counts[cls] > 0: alert_flag = True alert_reason.append((cls, class_counts[cls])) if class_counts.get(0, 0) > 5: alert_flag = True alert_reason.append((0, class_counts[0])) text = 'ALERT!' font = cv2.FONT_HERSHEY_DUPLEX font_scale = 0.75 thickness = 2 size = cv2.getTextSize(text, font, font_scale, thickness) x = 0 y = int((2 + size[0][1])) img = results[0].plot() if alert_flag: # Resize the red-tinted overlay to match the image size red_tinted_overlay = cv2.resize(red_tinted_overlay, (img.shape[1], img.shape[0])) img = cv2.addWeighted(img, 0.7, red_tinted_overlay, 0.3, 0) cv2.putText(img, text, (x, y), font, font_scale, (0, 0, 0), thickness) y += int(size[0][1]) + 10 # Move to the next line for cls, count in alert_reason: alert_text = f'{count} {alerting_classes[cls]}' cv2.putText(img, alert_text, (x, y), font, font_scale, (0, 0, 0), thickness) y += int(size[0][1]) + 10 # Move to the next line # Append the frame to the list frames.append(img) # Update processing progress bar current_frame_processing = int(cap.get(cv2.CAP_PROP_POS_FRAMES)) progress_bar_processing.progress(current_frame_processing / total_frames) frame_counter += 1 # Increment frame counter # Get the fps from the video capture object fps = cap.get(cv2.CAP_PROP_FPS) frame_delay = 1 / fps if fps > 0 else 1 / 24 # Use 24 fps as a fallback if fps is not available # Release resources del results cap.release() tfile.close() # Display frames one by one as a video for i, frame in enumerate(frames): video_placeholder.image(frame, channels="BGR", caption="YOLOv8 Inference") time.sleep(frame_delay) st.markdown("
", unsafe_allow_html=True) st.markdown(centered_text, unsafe_allow_html=True)