Update app.py

app.py

@@ -1,102 +1,27 @@
-import os
-for dirname, _, filenames in os.walk('/kaggle/input'):
-    for filename in filenames:
-        print(os.path.join(dirname, filename))
-
-
-
-
-#Importing Libraries
-import cv2
-import matplotlib.pyplot as plt
-%matplotlib inline
-from IPython.display import Image
-
-import keras_cv
-import keras_core as keras
-
-
-
-from collections import defaultdict
-import json
+from huggingface_hub import hf_hub_download
+from transformers import AutoImageProcessor, TableTransformerForObjectDetection
+import torch
+from PIL import Image
 
-class COCOParser:
-    def __init__(self, anns_file, imgs_dir):
-        with open(anns_file, 'r') as f:
-            coco = json.load(f)
-            
-        self.annIm_dict = defaultdict(list)        
-        self.cat_dict = {} 
-        self.annId_dict = {}
-        self.im_dict = {}
-        self.licenses_dict = {}
-       
-        for ann in coco['annotations']:           
-            self.annIm_dict[ann['image_id']].append(ann) 
-            self.annId_dict[ann['id']]=ann
-        for img in coco['images']:
-            self.im_dict[img['id']] = img
-        for cat in coco['categories']:
-            self.cat_dict[cat['id']] = cat
-        for license in coco['licenses']:
-            self.licenses_dict[license['id']] = license
-    
-    def get_imgIds(self):
-        return list(self.im_dict.keys())
-    def get_annIds(self, im_ids):
-        im_ids=im_ids if isinstance(im_ids, list) else [im_ids]
-        return [ann['id'] for im_id in im_ids for ann in self.annIm_dict[im_id]]
-    def load_anns(self, ann_ids):
-        im_ids=ann_ids if isinstance(ann_ids, list) else [ann_ids]
-        return [self.annId_dict[ann_id] for ann_id in ann_ids]        
-    def load_cats(self, class_ids):
-        class_ids=class_ids if isinstance(class_ids, list) else [class_ids]
-        return [self.cat_dict[class_id] for class_id in class_ids]
-    def get_imgLicenses(self,im_ids):
-        im_ids=im_ids if isinstance(im_ids, list) else [im_ids]
-        lic_ids = [self.im_dict[im_id]["license"] for im_id in im_ids]
-        return [self.licenses_dict[lic_id] for lic_id in lic_ids]
-coco_images_dir = "/kaggle/input/coco-2017-dataset/coco2017/train2017"
-annot_file = "/kaggle/input/coco-2017-dataset/coco2017/annotations/instances_train2017.json"
+file_path = hf_hub_download(repo_id="nielsr/example-pdf", repo_type="dataset", filename="example_pdf.png")
+image = Image.open(file_path).convert("RGB")
 
-coco = COCOParser(annot_file, coco_images_dir)
-from PIL import Image
-import numpy as np
+image_processor = AutoImageProcessor.from_pretrained("microsoft/table-transformer-detection")
+model = TableTransformerForObjectDetection.from_pretrained("microsoft/table-transformer-detection")
 
-# define a list of colors for drawing bounding boxes
-color_list = ["pink", "red", "teal", "blue", "orange", "yellow", "black", "magenta","green","aqua"]*10
-num_imgs_to_disp = 4
-total_images = len(coco.get_imgIds()) # total number of images
-sel_im_idxs = np.random.permutation(total_images)[:num_imgs_to_disp]
-img_ids = coco.get_imgIds()
-selected_img_ids = [img_ids[i] for i in sel_im_idxs]
-ann_ids = coco.get_annIds(selected_img_ids)
-im_licenses = coco.get_imgLicenses(selected_img_ids)
+inputs = image_processor(images=image, return_tensors="pt")
+outputs = model(**inputs)
 
-fig, ax = plt.subplots(nrows=2, ncols=2, figsize=(15,10))
-ax = ax.ravel()
+# convert outputs (bounding boxes and class logits) to Pascal VOC format (xmin, ymin, xmax, ymax)
+target_sizes = torch.tensor([image.size[::-1]])
+results = image_processor.post_process_object_detection(outputs, threshold=0.9, target_sizes=target_sizes)[
+    0
+]
 
-#Loading and visualizing the dataset
-for i, im in enumerate(selected_img_ids):
-    image = Image.open(f"{coco_images_dir}/{str(im).zfill(12)}.jpg")
-    ann_ids = coco.get_annIds(im)
-    annotations = coco.load_anns(ann_ids)
-    
-    for ann in annotations:
-        bbox = ann['bbox']
-        x, y, w, h = [int(b) for b in bbox]
-        class_id = ann["category_id"]
-        class_name = coco.load_cats(class_id)[0]["name"]
-        license = coco.get_imgLicenses(im)[0]["name"]
-        color_ = color_list[class_id]
-        rect = plt.Rectangle((x, y), w, h, linewidth=2, edgecolor=color_, facecolor='none')
-        t_box=ax[i].text(x, y, class_name,  color='red', fontsize=10)
-        t_box.set_bbox(dict(boxstyle='square, pad=0',facecolor='white', alpha=0.6, edgecolor='blue'))
-        ax[i].add_patch(rect)
-    
-    ax[i].axis('off')
-    ax[i].imshow(image)
-    ax[i].set_xlabel('Longitude')
-    ax[i].set_title(f"License: {license}")
-plt.tight_layout()
-plt.show()
+for score, label, box in zip(results["scores"], results["labels"], results["boxes"]):
+    box = [round(i, 2) for i in box.tolist()]
+    print(
+        f"Detected {model.config.id2label[label.item()]} with confidence "
+        f"{round(score.item(), 3)} at location {box}"
+    )
+Detected table with confidence 1.0 at location [202.1, 210.59, 1119.22, 385.09]