Update app.py

app.py

@@ -1,6 +1,5 @@
 import gradio as gr
 import torch
-import cv2
 import pytesseract
 from transformers import AutoTokenizer, AutoModelForSequenceClassification
 
@@ -13,12 +12,10 @@ classification_model_name = "distilbert-base-uncased"
 classification_tokenizer = AutoTokenizer.from_pretrained(classification_model_name)
 classification_model = AutoModelForSequenceClassification.from_pretrained(classification_model_name)
 
-# Define function for OCR
+# Define function for text recognition (OCR)
 def perform_ocr(image):
-    # Convert image to grayscale
-    gray_image = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)
-    # Perform OCR
-    text = pytesseract.image_to_string(gray_image)
+    # Perform OCR using pytesseract
+    text = pytesseract.image_to_string(image)
     return text
 
 # Define function to process video and predict
@@ -36,15 +33,12 @@ def predict_runner_status(video_file):
         results_detection = model(frame)
 
         # Logic for determining runner status using detected objects
-        # Example: if person detected, extract text and classify
         objects = results_detection.pred[0][:, -1].numpy()
         if 0 in objects:  # 0 corresponds to person class
-            # Get the cropped region containing the person for OCR
+            # Perform OCR on the detected person
             person_bbox = results_detection.pred[0][np.where(objects == 0)][0][:4]
             person_bbox = person_bbox.astype(int)
             person_img = frame[person_bbox[1]:person_bbox[3], person_bbox[0]:person_bbox[2]]
-
-            # Perform OCR on the cropped image
             text = perform_ocr(person_img)
 
             # Classification using text classification model