{
  "id": 181003,
  "title": "Batch inference in DELG",
  "url": "/competitions/landmark-recognition-2020/discussion/181003",
  "author_name": "Vladislav Ostankovich",
  "post_date": "2020-09-07T08:30:56.711000",
  "votes": 9,
  "comment_count": 0,
  "views": 0,
  "content": "<p>Hi, I wonder if DELG is able to perform batch inference? In DELG repo on github and baseline notebooks on Kaggle, the inference is always done on a single image via extract_local_features() that works with single tensor of shape (height, width, 3). Is there a way to pass something like numpy.array or tf.tensor of shape (batch_size, height, width, 3) to the DELG model? Thank you! </p>",
  "messages": [
    {
      "id": 1001328,
      "postDate": "2020-09-07T08:30:56.713Z",
      "content": "<p>Hi, I wonder if DELG is able to perform batch inference? In DELG repo on github and baseline notebooks on Kaggle, the inference is always done on a single image via extract_local_features() that works with single tensor of shape (height, width, 3). Is there a way to pass something like numpy.array or tf.tensor of shape (batch_size, height, width, 3) to the DELG model? Thank you! </p>",
      "rawMarkdown": "Hi, I wonder if DELG is able to perform batch inference? In DELG repo on github and baseline notebooks on Kaggle, the inference is always done on a single image via extract_local_features() that works with single tensor of shape (height, width, 3). Is there a way to pass something like numpy.array or tf.tensor of shape (batch_size, height, width, 3) to the DELG model? Thank you! ",
      "votes": 9
    }
  ],
  "comments": [],
  "raw_markdown_by_id": {
    "1001328": "Hi, I wonder if DELG is able to perform batch inference? In DELG repo on github and baseline notebooks on Kaggle, the inference is always done on a single image via extract_local_features() that works with single tensor of shape (height, width, 3). Is there a way to pass something like numpy.array or tf.tensor of shape (batch_size, height, width, 3) to the DELG model? Thank you! "
  }
}