Spaces:
Sleeping
Sleeping
| """ | |
| Reference | |
| - https://docs.streamlit.io/library/api-reference/layout | |
| - https://github.com/CodingMantras/yolov8-streamlit-detection-tracking/blob/master/app.py | |
| - https://huggingface.co/keremberke/yolov8m-valorant-detection/tree/main | |
| - https://docs.ultralytics.com/usage/python/ | |
| """ | |
| import time | |
| import PIL | |
| import streamlit as st | |
| import torch | |
| from ultralyticsplus import YOLO, render_result | |
| from convert import convert_to_braille_unicode, parse_xywh_and_class, braille_to_text | |
| def load_model(model_path): | |
| """load model from path""" | |
| model = YOLO(model_path) | |
| return model | |
| def load_image(image_path): | |
| """load image from path""" | |
| image = PIL.Image.open(image_path) | |
| return image | |
| # title | |
| st.title("Braille Pattern Detection") | |
| # sidebar | |
| st.sidebar.header("Detection Config") | |
| conf = float(st.sidebar.slider("Class Confidence", 10, 75, 15)) / 100 | |
| iou = float(st.sidebar.slider("IoU Threshold", 10, 75, 15)) / 100 | |
| model_path = "snoop2head/yolov8m-braille" | |
| try: | |
| model = load_model(model_path) | |
| model.overrides["conf"] = conf # NMS confidence threshold | |
| model.overrides["iou"] = iou # NMS IoU threshold | |
| model.overrides["agnostic_nms"] = False # NMS class-agnostic | |
| model.overrides["max_det"] = 1000 # maximum number of detections per image | |
| except Exception as ex: | |
| print(ex) | |
| st.write(f"Unable to load model. Check the specified path: {model_path}") | |
| source_img = None | |
| source_img = st.sidebar.file_uploader( | |
| "Choose an image...", type=("jpg", "jpeg", "png", "bmp", "webp") | |
| ) | |
| # Single column layout - only show uploaded image | |
| if source_img is None: | |
| default_image_path = "./images/alpha-numeric.jpeg" | |
| image = load_image(default_image_path) | |
| st.image( | |
| default_image_path, caption="Example Input Image", use_column_width=True | |
| ) | |
| else: | |
| image = load_image(source_img) | |
| st.image(source_img, caption="Uploaded Image", use_column_width=True) | |
| # Process the image | |
| with st.spinner("Wait for it..."): | |
| start_time = time.time() | |
| try: | |
| with torch.no_grad(): | |
| res = model.predict( | |
| image, save=True, save_txt=True, exist_ok=True, conf=conf | |
| ) | |
| boxes = res[0].boxes # first image | |
| list_boxes = parse_xywh_and_class(boxes) | |
| except Exception as ex: | |
| st.write("Please upload image with types of JPG, JPEG, PNG ...") | |
| # Convert braille to text and display results | |
| detected_text_lines = [] | |
| try: | |
| st.success(f"Done! Inference time: {time.time() - start_time:.2f} seconds") | |
| st.subheader("Detected Text") | |
| for box_line in list_boxes: | |
| str_left_to_right = "" | |
| box_classes = box_line[:, -1] | |
| for each_class in box_classes: | |
| braille_unicode = convert_to_braille_unicode(model.names[int(each_class)]) | |
| str_left_to_right += braille_unicode | |
| # Convert braille unicode to actual text | |
| text_line = braille_to_text(str_left_to_right) | |
| detected_text_lines.append(text_line) | |
| st.write(text_line) | |
| # Combine all detected text | |
| full_detected_text = "\n".join(detected_text_lines) | |
| # Add copy to clipboard functionality | |
| if st.button("Copy Text to Clipboard"): | |
| # Use streamlit's built-in clipboard functionality | |
| st.code(full_detected_text, language=None) | |
| st.success("Text displayed above. Use Ctrl+A to select all, then Ctrl+C to copy!") | |
| # Alternative: Create a text area that users can easily copy from | |
| st.subheader("Copy Text Below:") | |
| st.text_area("Detected Text (Select All & Copy)", value=full_detected_text, height=150) | |
| except Exception as ex: | |
| st.write("Please try again with images with types of JPG, JPEG, PNG ...") |