From f01e549e6189f1d2db3f0d2ab7cee565ffef65de Mon Sep 17 00:00:00 2001
From: onb1259 <onb1259@onb.ac.at>
Date: Mon, 13 Nov 2023 11:27:39 +0100
Subject: [PATCH] added functionality to add iiif annotations to manifest
---
extract_figures.ipynb | 13 ++++++++++---
iiif_utils.py | 43 +++++++++++++++++++++++++++++++++----------
2 files changed, 43 insertions(+), 13 deletions(-)
diff --git a/extract_figures.ipynb b/extract_figures.ipynb
index 13bdbd8..b305145 100644
--- a/extract_figures.ipynb
+++ b/extract_figures.ipynb
@@ -21,12 +21,13 @@
"outputs": [],
"source": [
"import os, shutil\n",
+ "import json\n",
"from zipapp import zipfile\n",
"from PIL import Image\n",
"from ultralytics import YOLO\n",
"from IPython.display import display, FileLink, HTML\n",
"\n",
- "from iiif_utils import get_imgurls_from_manifesturl, create_paths_from_iiifurls, download_images_multithreded\n",
+ "from iiif_utils import get_imgurls_from_manifesturl, create_paths_from_iiifurls, download_images_multithreded, add_annotations, get_json_by_url\n",
"\n",
"# Url of iiif manifest\n",
"input_manifest = 'https://iiif.onb.ac.at/presentation/ABO/+Z105572305/manifest'"
@@ -69,17 +70,23 @@
"os.mkdir(RESULTS_DIR)\n",
"model = YOLO('model_extract_figures.pt')\n",
"counter = 0\n",
- "for path in imgpaths:\n",
+ "list_of_bbs = [None] * len(imgpaths)\n",
+ "for j, path in enumerate(imgpaths):\n",
" result = model.predict(path, verbose=False)\n",
" img = Image.open(path)\n",
" for i, bb in enumerate(result[0].boxes.xyxy.tolist()):\n",
" #print(f'Figure detected in: {path}')\n",
" counter += 1\n",
" bb = [int(e) for e in bb]\n",
+ " list_of_bbs[j] = bb\n",
" img_cropped = img.crop(bb)\n",
" path_cropped = os.path.join(RESULTS_DIR, os.path.basename(path.rsplit('.', 1)[0] + '_crop' + str(i) + '.' + path.rsplit('.', 1)[1]))\n",
" img_cropped.save(path_cropped)\n",
- "print(f'Total number of extracted figures: {counter}')"
+ "print(f'Total number of extracted figures: {counter}')\n",
+ "\n",
+ "manifest_annotated = add_annotations(list_of_bbs, get_json_by_url(input_manifest))\n",
+ "with open(os.path.join(RESULTS_DIR, 'annotated_manifest.json'), 'w+') as f:\n",
+ " json.dump(manifest_annotated, f, indent=2)"
]
},
{
diff --git a/iiif_utils.py b/iiif_utils.py
index 723c2e8..69418c9 100644
--- a/iiif_utils.py
+++ b/iiif_utils.py
@@ -56,14 +56,37 @@ def create_paths_from_iiifurls(img_urls, base_path=''):
res = [os.path.join(base_path, el) for el in res]
return res
+def add_annotations(list_of_boundingboxes, manifest):
+ for i, _ in enumerate(manifest['sequences'][0]['canvases']):
+ id = manifest['sequences'][0]['canvases'][i]['images'][0]['@id']
+ bounding_box = list_of_boundingboxes[i]
+ # if bounding box is None, skip
+ if not bounding_box:
+ continue
+ annotation = {
+ "@context": "http://iiif.io/api/presentation/2/context.json",
+ "@type": "sc:AnnotationList",
+ "@id": f"{id}/annotations",
+ "resources": [
+ {
+ "@type": "oa:Annotation",
+ "motivation": "sc:painting",
+ "resource": {
+ "@type": "oa:Choice",
+ "default": {
+ "@type": "oa:SpecificResource",
+ "selector": {
+ "@type": "oa:FragmentSelector",
+ "value": f"xywh={bounding_box[0]},{bounding_box[1]},{bounding_box[0]+bounding_box[2]},{bounding_box[2]+bounding_box[3]}"
+ },
+ "style": "rect",
+ "label": "Detected figure"
+ }
+ },
+ "on": f"{id}"
+ }
+ ]
+ }
+ manifest['sequences'][0]['canvases'][i]['otherContent'].append(annotation)
+ return manifest
-# collection_url = 'https://iiif.onb.ac.at/presentation/collection/labs_botanical_illustrations'
-# urls = get_imgurls_from_collectionurl(collection_url)
-# paths = create_paths_from_iiifurls(urls, 'downloads')
-# download_images_multithreded(urls, paths)
-
-# collection_url = 'https://iiif.onb.ac.at/presentation/collection/apz_1841'
-# manifest_url = 'https://iiif.onb.ac.at/presentation/ANNO/apz18411229/manifest/'
-# urls = get_imgurls_from_manifesturl(manifest_url)
-# paths = create_paths_from_iiifurls(urls, 'downloads')
-# download_images_multithreded(urls, paths)
--
GitLab