ধারণাগত_ক্যাপশন

তথ্যসূত্র:

TFDS এ এই ডেটাসেট লোড করতে নিম্নলিখিত কমান্ডটি ব্যবহার করুন:

ds = tfds.load('huggingface:conceptual_captions')
  • বর্ণনা :
Image captioning dataset
The resulting dataset (version 1.1) has been split into Training, Validation, and Test splits. The Training split consists of 3,318,333 image-URL/caption pairs, with a total number of 51,201 total token types in the captions (i.e., total vocabulary). The average number of tokens per captions is 10.3 (standard deviation of 4.5), while the median is 9.0 tokens per caption. The Validation split consists of 15,840 image-URL/caption pairs, with similar statistics.
  • লাইসেন্স : কোনো পরিচিত লাইসেন্স নেই
  • সংস্করণ : 1.1.0
  • বিভাজন :
বিভক্ত উদাহরণ
'train' 3318333
'validation' 15840
  • বৈশিষ্ট্য :
{
    "id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "caption": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "url": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    }
}

লেবেলবিহীন

TFDS এ এই ডেটাসেট লোড করতে নিম্নলিখিত কমান্ডটি ব্যবহার করুন:

ds = tfds.load('huggingface:conceptual_captions/unlabeled')
  • বর্ণনা :
Google's Conceptual Captions dataset has more than 3 million images, paired with natural-language captions.
In contrast with the curated style of the MS-COCO images, Conceptual Captions images and their raw descriptions are harvested from the web,
and therefore represent a wider variety of styles. The raw descriptions are harvested from the Alt-text HTML attribute associated with web images.
The authors developed an automatic pipeline that extracts, filters, and transforms candidate image/caption pairs, with the goal of achieving a balance of cleanliness,
informativeness, fluency, and learnability of the resulting captions.
  • লাইসেন্স : ডেটাসেটটি যেকোন উদ্দেশ্যে অবাধে ব্যবহার করা যেতে পারে, যদিও ডেটা উৎস হিসেবে Google LLC ("Google") এর স্বীকৃতির প্রশংসা করা হবে। ডেটাসেটটি "AS IS" কোনো ওয়ারেন্টি, প্রকাশ বা উহ্য ছাড়াই প্রদান করা হয়েছে। ডেটাসেট ব্যবহারের ফলে প্রত্যক্ষ বা পরোক্ষভাবে যে কোনো ক্ষতির জন্য Google সমস্ত দায় অস্বীকার করে।

  • সংস্করণ : 0.0.0

  • বিভাজন :

বিভক্ত উদাহরণ
'train' 3318333
'validation' 15840
  • বৈশিষ্ট্য :
{
    "image_url": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "caption": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    }
}

লেবেলযুক্ত

TFDS এ এই ডেটাসেট লোড করতে নিম্নলিখিত কমান্ডটি ব্যবহার করুন:

ds = tfds.load('huggingface:conceptual_captions/labeled')
  • বর্ণনা :
Google's Conceptual Captions dataset has more than 3 million images, paired with natural-language captions.
In contrast with the curated style of the MS-COCO images, Conceptual Captions images and their raw descriptions are harvested from the web,
and therefore represent a wider variety of styles. The raw descriptions are harvested from the Alt-text HTML attribute associated with web images.
The authors developed an automatic pipeline that extracts, filters, and transforms candidate image/caption pairs, with the goal of achieving a balance of cleanliness,
informativeness, fluency, and learnability of the resulting captions.
  • লাইসেন্স : ডেটাসেটটি যেকোন উদ্দেশ্যে অবাধে ব্যবহার করা যেতে পারে, যদিও ডেটা উৎস হিসেবে Google LLC ("Google") এর স্বীকৃতির প্রশংসা করা হবে। ডেটাসেটটি "AS IS" কোনো ওয়ারেন্টি, প্রকাশ বা উহ্য ছাড়াই প্রদান করা হয়েছে। ডেটাসেট ব্যবহারের ফলে প্রত্যক্ষ বা পরোক্ষভাবে যে কোনো ক্ষতির জন্য Google সমস্ত দায় অস্বীকার করে।

  • সংস্করণ : 0.0.0

  • বিভাজন :

বিভক্ত উদাহরণ
'train' 2007090
  • বৈশিষ্ট্য :
{
    "image_url": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "caption": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "labels": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "MIDs": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "confidence_scores": {
        "feature": {
            "dtype": "float64",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    }
}