input stringlengths 22 960 | output unknown | schema unknown |
|---|---|---|
We propose CornerNet , a new approach to object detection where we detect an object bounding box as a pair of keypoints , the top - left corner and the bottom - right corner , using a single convolution neural network . | {
"entities": {
"method": [
{
"text": "CornerNet",
"start": 11,
"end": 20
},
{
"text": "convolution neural network",
"start": 191,
"end": 217
}
],
"task": [
{
"text": "object detection",
"start": 41,
"end... | {
"entities": [
{
"label": "dataset",
"description": "A realistic collection of data that is used for training, validating, or testing the algorithms. These datasets can consist of various forms of data such as text, images, videos, or structured data. For example, MNIST, COCO, AGNews, IMDb, etc.",
... |
Experiments show that CornerNet achieves a 4 2 . 2 % AP on MS COCO , outperforming all existing one - stage detectors . | {
"entities": {
"method": [
{
"text": "CornerNet",
"start": 22,
"end": 31
}
],
"dataset": [
{
"text": "MS COCO",
"start": 59,
"end": 66
}
]
},
"relations": {
"evaluated with": [
{
"head": {
"text": ... | {
"entities": [
{
"label": "dataset",
"description": "A realistic collection of data that is used for training, validating, or testing the algorithms. These datasets can consist of various forms of data such as text, images, videos, or structured data. For example, MNIST, COCO, AGNews, IMDb, etc.",
... |
Object detectors based on convolutional neural networks ( ConvNets ) ( Krizhevsky et al. , 2 0 1 2 ; Simonyan and Zisserman , 2 0 1 4 ; He et al. , 2 0 1 6 ) have achieved state - of - the - art results on various challenging benchmarks ( Lin et al. , 2 0 1 4 ; Deng et al. , 2 0 0 9 ; Everingham et al. , 2 0 1 5 ) . | {
"entities": {
"method": [
{
"text": "convolutional neural networks",
"start": 26,
"end": 55
},
{
"text": "ConvNets",
"start": 58,
"end": 66
}
]
},
"relations": {
"synonym of": [
{
"head": {
"text": "ConvN... | {
"entities": [
{
"label": "dataset",
"description": "A realistic collection of data that is used for training, validating, or testing the algorithms. These datasets can consist of various forms of data such as text, images, videos, or structured data. For example, MNIST, COCO, AGNews, IMDb, etc.",
... |
In this paper we introduce CornerNet , a new onestage approach to object detection that does away with anchor boxes . | {
"entities": {
"method": [
{
"text": "CornerNet",
"start": 27,
"end": 36
}
],
"task": [
{
"text": "object detection",
"start": 66,
"end": 82
}
]
},
"relations": {
"used for": [
{
"head": {
"text": ... | {
"entities": [
{
"label": "dataset",
"description": "A realistic collection of data that is used for training, validating, or testing the algorithms. These datasets can consist of various forms of data such as text, images, videos, or structured data. For example, MNIST, COCO, AGNews, IMDb, etc.",
... |
We use a single convolutional network to predict a heatmap for the top - left corners of all instances of the same object category , a heatmap for all bottomright corners , and an embedding vector for each detected corner . | {
"entities": {
"method": [
{
"text": "convolutional network",
"start": 16,
"end": 37
}
]
},
"relations": {}
} | {
"entities": [
{
"label": "dataset",
"description": "A realistic collection of data that is used for training, validating, or testing the algorithms. These datasets can consist of various forms of data such as text, images, videos, or structured data. For example, MNIST, COCO, AGNews, IMDb, etc.",
... |
A convolutional network outputs a heatmap for all top - left corners , a heatmap for all bottom - right corners , and an embedding vector for each detected corner . | {
"entities": {
"method": [
{
"text": "convolutional network",
"start": 2,
"end": 23
}
]
},
"relations": {}
} | {
"entities": [
{
"label": "dataset",
"description": "A realistic collection of data that is used for training, validating, or testing the algorithms. These datasets can consist of various forms of data such as text, images, videos, or structured data. For example, MNIST, COCO, AGNews, IMDb, etc.",
... |
Another novel component of CornerNet is corner pooling , a new type of pooling layer that helps a convolutional network better localize corners of bounding boxes . | {
"entities": {
"method": [
{
"text": "CornerNet",
"start": 27,
"end": 36
},
{
"text": "corner pooling",
"start": 40,
"end": 54
}
]
},
"relations": {
"part of": [
{
"head": {
"text": "corner pooling",
... | {
"entities": [
{
"label": "dataset",
"description": "A realistic collection of data that is used for training, validating, or testing the algorithms. These datasets can consist of various forms of data such as text, images, videos, or structured data. For example, MNIST, COCO, AGNews, IMDb, etc.",
... |
We demonstrate the effectiveness of CornerNet on MS COCO ( Lin et al. , 2 0 1 4 ) . | {
"entities": {
"method": [
{
"text": "CornerNet",
"start": 36,
"end": 45
}
],
"dataset": [
{
"text": "MS COCO",
"start": 49,
"end": 56
}
]
},
"relations": {
"used for": [
{
"head": {
"text": "Corne... | {
"entities": [
{
"label": "dataset",
"description": "A realistic collection of data that is used for training, validating, or testing the algorithms. These datasets can consist of various forms of data such as text, images, videos, or structured data. For example, MNIST, COCO, AGNews, IMDb, etc.",
... |
In addition , through ablation studies we show that corner pooling is critical to the superior performance of CornerNet . | {
"entities": {
"method": [
{
"text": "corner pooling",
"start": 52,
"end": 66
},
{
"text": "CornerNet",
"start": 110,
"end": 119
}
]
},
"relations": {
"part of": [
{
"head": {
"text": "corner pooling",
... | {
"entities": [
{
"label": "dataset",
"description": "A realistic collection of data that is used for training, validating, or testing the algorithms. These datasets can consist of various forms of data such as text, images, videos, or structured data. For example, MNIST, COCO, AGNews, IMDb, etc.",
... |
Code is available at https://github.com/ princeton - vl/CornerNet . Two - stage approach was first introduced and popularized by R - CNN ( Girshick et al. , 2 0 1 4 ) . | {
"entities": {
"method": [
{
"text": "vl/CornerNet",
"start": 53,
"end": 65
},
{
"text": "R - CNN",
"start": 131,
"end": 138
}
]
},
"relations": {}
} | {
"entities": [
{
"label": "dataset",
"description": "A realistic collection of data that is used for training, validating, or testing the algorithms. These datasets can consist of various forms of data such as text, images, videos, or structured data. For example, MNIST, COCO, AGNews, IMDb, etc.",
... |
Later , SPP ( He et al. , 2 0 1 4 ) and Fast - RCNN ( Girshick , 2 0 1 5 ) improve R - CNN by designing a special pooling layer that pools each region from feature maps instead . | {
"entities": {
"method": [
{
"text": "SPP",
"start": 8,
"end": 11
},
{
"text": "Fast - RCNN",
"start": 40,
"end": 51
},
{
"text": "R - CNN",
"start": 83,
"end": 90
}
]
},
"relations": {
"compar... | {
"entities": [
{
"label": "dataset",
"description": "A realistic collection of data that is used for training, validating, or testing the algorithms. These datasets can consist of various forms of data such as text, images, videos, or structured data. For example, MNIST, COCO, AGNews, IMDb, etc.",
... |
Faster - RCNN ( Ren et al. , 2 0 1 5 ) does away low level proposal algorithms by introducing a region proposal network ( RPN ) , which generates proposals from a set of pre - determined candidate boxes , usually known as anchor boxes . | {
"entities": {
"method": [
{
"text": "Faster - RCNN",
"start": 0,
"end": 13
},
{
"text": "region proposal network",
"start": 96,
"end": 119
},
{
"text": "RPN",
"start": 122,
"end": 125
}
]
},
"rela... | {
"entities": [
{
"label": "dataset",
"description": "A realistic collection of data that is used for training, validating, or testing the algorithms. These datasets can consist of various forms of data such as text, images, videos, or structured data. For example, MNIST, COCO, AGNews, IMDb, etc.",
... |
R - FCN ( Dai et al. , 2 0 1 6 ) further improves the efficiency of Faster - RCNN by replacing the fully connected sub - detection network with a fully convolutional subdetection network . | {
"entities": {
"method": [
{
"text": "R - FCN",
"start": 0,
"end": 7
},
{
"text": "Faster - RCNN",
"start": 68,
"end": 81
},
{
"text": "fully connected sub - detection network",
"start": 99,
"end": 138
},
... | {
"entities": [
{
"label": "dataset",
"description": "A realistic collection of data that is used for training, validating, or testing the algorithms. These datasets can consist of various forms of data such as text, images, videos, or structured data. For example, MNIST, COCO, AGNews, IMDb, etc.",
... |
On the other hand , YOLO and SSD ( Liu et al. , 2 0 1 6 ) have popularized the one - stage approach , which removes the RoI pooling step and detects objects in a single network . | {
"entities": {
"method": [
{
"text": "YOLO",
"start": 20,
"end": 24
},
{
"text": "SSD",
"start": 29,
"end": 32
}
]
},
"relations": {}
} | {
"entities": [
{
"label": "dataset",
"description": "A realistic collection of data that is used for training, validating, or testing the algorithms. These datasets can consist of various forms of data such as text, images, videos, or structured data. For example, MNIST, COCO, AGNews, IMDb, etc.",
... |
YOLO predicts bounding box coordinates directly from an image , and is later improved in YOLO 9 0 0 0 by switching to anchor boxes . | {
"entities": {
"method": [
{
"text": "YOLO",
"start": 0,
"end": 4
},
{
"text": "YOLO 9 0 0 0",
"start": 89,
"end": 101
}
]
},
"relations": {
"compare with": [
{
"head": {
"text": "YOLO",
"start":... | {
"entities": [
{
"label": "dataset",
"description": "A realistic collection of data that is used for training, validating, or testing the algorithms. These datasets can consist of various forms of data such as text, images, videos, or structured data. For example, MNIST, COCO, AGNews, IMDb, etc.",
... |
DSSD and RON ( Kong et al. , 2 0 1 7 ) adopt networks similar to the hourglass network ( Newell et al. , 2 0 1 6 ) , enabling them to combine low - level and high - level features via skip connections to predict bounding boxes more accurately . | {
"entities": {
"method": [
{
"text": "DSSD",
"start": 0,
"end": 4
},
{
"text": "RON",
"start": 9,
"end": 12
},
{
"text": "hourglass network",
"start": 69,
"end": 86
},
{
"text": "skip connect... | {
"entities": [
{
"label": "dataset",
"description": "A realistic collection of data that is used for training, validating, or testing the algorithms. These datasets can consist of various forms of data such as text, images, videos, or structured data. For example, MNIST, COCO, AGNews, IMDb, etc.",
... |
Second , DeNet selects features at manually determined locations relative to a region for classification , while our approach does not require any feature selection step . | {
"entities": {
"method": [
{
"text": "DeNet",
"start": 9,
"end": 14
}
],
"task": [
{
"text": "classification",
"start": 90,
"end": 104
},
{
"text": "feature selection",
"start": 147,
"end": 164
}
... | {
"entities": [
{
"label": "dataset",
"description": "A realistic collection of data that is used for training, validating, or testing the algorithms. These datasets can consist of various forms of data such as text, images, videos, or structured data. For example, MNIST, COCO, AGNews, IMDb, etc.",
... |
Point Linking Network ( PLN ) is an one - stage detector without anchor boxes . | {
"entities": {
"method": [
{
"text": "Point Linking Network",
"start": 0,
"end": 21
},
{
"text": "PLN",
"start": 24,
"end": 27
}
]
},
"relations": {
"synonym of": [
{
"head": {
"text": "PLN",
"st... | {
"entities": [
{
"label": "dataset",
"description": "A realistic collection of data that is used for training, validating, or testing the algorithms. These datasets can consist of various forms of data such as text, images, videos, or structured data. For example, MNIST, COCO, AGNews, IMDb, etc.",
... |
CornerNet is very different from PLN . | {
"entities": {
"method": [
{
"text": "CornerNet",
"start": 0,
"end": 9
},
{
"text": "PLN",
"start": 33,
"end": 36
}
]
},
"relations": {
"compare with": [
{
"head": {
"text": "CornerNet",
"start":... | {
"entities": [
{
"label": "dataset",
"description": "A realistic collection of data that is used for training, validating, or testing the algorithms. These datasets can consist of various forms of data such as text, images, videos, or structured data. For example, MNIST, COCO, AGNews, IMDb, etc.",
... |
Second , CornerNet uses corner pooling to better localize the corners . | {
"entities": {
"method": [
{
"text": "CornerNet",
"start": 9,
"end": 18
},
{
"text": "corner pooling",
"start": 24,
"end": 38
}
]
},
"relations": {
"part of": [
{
"head": {
"text": "corner pooling",
... | {
"entities": [
{
"label": "dataset",
"description": "A realistic collection of data that is used for training, validating, or testing the algorithms. These datasets can consist of various forms of data such as text, images, videos, or structured data. For example, MNIST, COCO, AGNews, IMDb, etc.",
... |
We also significantly modify the hourglass architecture and add our novel variant of focal loss to help better train the network . In CornerNet , we detect an object as a pair of keypointsthe top - left corner and bottom - right corner of the bounding box . | {
"entities": {
"method": [
{
"text": "focal loss",
"start": 85,
"end": 95
},
{
"text": "CornerNet",
"start": 136,
"end": 145
}
]
},
"relations": {}
} | {
"entities": [
{
"label": "dataset",
"description": "A realistic collection of data that is used for training, validating, or testing the algorithms. These datasets can consist of various forms of data such as text, images, videos, or structured data. For example, MNIST, COCO, AGNews, IMDb, etc.",
... |
We use the hourglass network ( Newell et al. , 2 0 1 6 ) as the backbone network of CornerNet . | {
"entities": {
"method": [
{
"text": "hourglass network",
"start": 11,
"end": 28
},
{
"text": "CornerNet",
"start": 84,
"end": 93
}
]
},
"relations": {
"part of": [
{
"head": {
"text": "hourglass network",... | {
"entities": [
{
"label": "dataset",
"description": "A realistic collection of data that is used for training, validating, or testing the algorithms. These datasets can consist of various forms of data such as text, images, videos, or structured data. For example, MNIST, COCO, AGNews, IMDb, etc.",
... |
Each module has its own corner pooling module to pool features from the hourglass network before predicting the heatmaps , embeddings and offsets . | {
"entities": {
"method": [
{
"text": "corner pooling",
"start": 24,
"end": 38
},
{
"text": "hourglass network",
"start": 72,
"end": 89
}
]
},
"relations": {}
} | {
"entities": [
{
"label": "dataset",
"description": "A realistic collection of data that is used for training, validating, or testing the algorithms. These datasets can consist of various forms of data such as text, images, videos, or structured data. For example, MNIST, COCO, AGNews, IMDb, etc.",
... |
In this modified residual block , we replace the first 3 × 3 convolution module with a corner pooling module , which first processes the features from the backbone network by two 3 × 3 convolution modules 1 with 1 2 8 channels and then applies a corner pooling layer . | {
"entities": {
"method": [
{
"text": "residual block",
"start": 17,
"end": 31
},
{
"text": "3 × 3 convolution",
"start": 55,
"end": 72
},
{
"text": "corner pooling",
"start": 87,
"end": 101
},
{
... | {
"entities": [
{
"label": "dataset",
"description": "A realistic collection of data that is used for training, validating, or testing the algorithms. These datasets can consist of various forms of data such as text, images, videos, or structured data. For example, MNIST, COCO, AGNews, IMDb, etc.",
... |
Following the design of a residual block , we then feed the pooled features into a 3 × 3 Conv - BN layer with 2 5 6 channels and add back the projection shortcut . | {
"entities": {
"method": [
{
"text": "residual block",
"start": 26,
"end": 40
},
{
"text": "3 × 3 Conv - BN",
"start": 83,
"end": 98
}
]
},
"relations": {}
} | {
"entities": [
{
"label": "dataset",
"description": "A realistic collection of data that is used for training, validating, or testing the algorithms. These datasets can consist of various forms of data such as text, images, videos, or structured data. For example, MNIST, COCO, AGNews, IMDb, etc.",
... |
The modified residual block is followed by a 3 × 3 convolution module with 2 5 6 channels , and 3 Conv - ReLU - Conv layers to produce the heatmaps , embeddings and offsets . | {
"entities": {
"method": [
{
"text": "residual block",
"start": 13,
"end": 27
},
{
"text": "3 × 3 convolution",
"start": 45,
"end": 62
},
{
"text": "Conv - ReLU - Conv",
"start": 98,
"end": 116
}
]
}... | {
"entities": [
{
"label": "dataset",
"description": "A realistic collection of data that is used for training, validating, or testing the algorithms. These datasets can consist of various forms of data such as text, images, videos, or structured data. For example, MNIST, COCO, AGNews, IMDb, etc.",
... |
CornerNet uses the hourglass network ( Newell et al. , 2 0 1 6 ) as its backbone network . | {
"entities": {
"method": [
{
"text": "CornerNet",
"start": 0,
"end": 9
},
{
"text": "hourglass network",
"start": 19,
"end": 36
}
]
},
"relations": {
"part of": [
{
"head": {
"text": "hourglass network",
... | {
"entities": [
{
"label": "dataset",
"description": "A realistic collection of data that is used for training, validating, or testing the algorithms. These datasets can consist of various forms of data such as text, images, videos, or structured data. For example, MNIST, COCO, AGNews, IMDb, etc.",
... |
The hourglass network was first introduced for the human pose estimation task . | {
"entities": {
"method": [
{
"text": "hourglass network",
"start": 4,
"end": 21
}
],
"task": [
{
"text": "human pose estimation",
"start": 51,
"end": 72
}
]
},
"relations": {
"used for": [
{
"head": {
... | {
"entities": [
{
"label": "dataset",
"description": "A realistic collection of data that is used for training, validating, or testing the algorithms. These datasets can consist of various forms of data such as text, images, videos, or structured data. For example, MNIST, COCO, AGNews, IMDb, etc.",
... |
An hourglass module first downsamples the input features by a series of convolution and max pooling layers . | {
"entities": {
"method": [
{
"text": "hourglass module",
"start": 3,
"end": 19
},
{
"text": "convolution",
"start": 72,
"end": 83
},
{
"text": "max pooling",
"start": 88,
"end": 99
}
]
},
"relation... | {
"entities": [
{
"label": "dataset",
"description": "A realistic collection of data that is used for training, validating, or testing the algorithms. These datasets can consist of various forms of data such as text, images, videos, or structured data. For example, MNIST, COCO, AGNews, IMDb, etc.",
... |
These properties make the hourglass network an ideal choice for object detection as well . | {
"entities": {
"method": [
{
"text": "hourglass network",
"start": 26,
"end": 43
}
],
"task": [
{
"text": "object detection",
"start": 64,
"end": 80
}
]
},
"relations": {
"used for": [
{
"head": {
... | {
"entities": [
{
"label": "dataset",
"description": "A realistic collection of data that is used for training, validating, or testing the algorithms. These datasets can consist of various forms of data such as text, images, videos, or structured data. For example, MNIST, COCO, AGNews, IMDb, etc.",
... |
Our hourglass network consists of two hourglasses , and we make some modifications to the architecture of the hourglass module . | {
"entities": {
"method": [
{
"text": "hourglass network",
"start": 4,
"end": 21
},
{
"text": "hourglasses",
"start": 38,
"end": 49
},
{
"text": "hourglass module",
"start": 110,
"end": 126
}
]
},
"... | {
"entities": [
{
"label": "dataset",
"description": "A realistic collection of data that is used for training, validating, or testing the algorithms. These datasets can consist of various forms of data such as text, images, videos, or structured data. For example, MNIST, COCO, AGNews, IMDb, etc.",
... |
We scan from right to left for the horizontal max - pooling and from bottom to top for the vertical max - pooling . | {
"entities": {
"method": [
{
"text": "horizontal max - pooling",
"start": 35,
"end": 59
},
{
"text": "vertical max - pooling",
"start": 91,
"end": 113
}
]
},
"relations": {}
} | {
"entities": [
{
"label": "dataset",
"description": "A realistic collection of data that is used for training, validating, or testing the algorithms. These datasets can consist of various forms of data such as text, images, videos, or structured data. For example, MNIST, COCO, AGNews, IMDb, etc.",
... |
Fig. 7 The prediction module starts with a modified residual block , in which we replace the first convolution module with our corner pooling module . | {
"entities": {
"method": [
{
"text": "residual block",
"start": 52,
"end": 66
},
{
"text": "convolution",
"start": 99,
"end": 110
},
{
"text": "corner pooling",
"start": 127,
"end": 141
}
]
},
"rel... | {
"entities": [
{
"label": "dataset",
"description": "A realistic collection of data that is used for training, validating, or testing the algorithms. These datasets can consist of various forms of data such as text, images, videos, or structured data. For example, MNIST, COCO, AGNews, IMDb, etc.",
... |
We apply a 1 × 1 Conv - BN module to both the input and output of the first hourglass module . | {
"entities": {
"method": [
{
"text": "1 × 1 Conv - BN",
"start": 11,
"end": 26
},
{
"text": "hourglass module",
"start": 76,
"end": 92
}
]
},
"relations": {
"part of": [
{
"head": {
"text": "1 × 1 Conv - B... | {
"entities": [
{
"label": "dataset",
"description": "A realistic collection of data that is used for training, validating, or testing the algorithms. These datasets can consist of various forms of data such as text, images, videos, or structured data. For example, MNIST, COCO, AGNews, IMDb, etc.",
... |
We then merge them by element - wise addition followed by a ReLU and a residual block with 2 5 6 channels , which is then used as the input to the second hourglass module . | {
"entities": {
"method": [
{
"text": "ReLU",
"start": 60,
"end": 64
},
{
"text": "residual block",
"start": 71,
"end": 85
},
{
"text": "hourglass module",
"start": 154,
"end": 170
}
]
},
"relations... | {
"entities": [
{
"label": "dataset",
"description": "A realistic collection of data that is used for training, validating, or testing the algorithms. These datasets can consist of various forms of data such as text, images, videos, or structured data. For example, MNIST, COCO, AGNews, IMDb, etc.",
... |
To reduce overfitting , we adopt standard data augmentation techniques including random horizontal flipping , random scaling , random cropping and random color jittering , which includes adjusting the brightness , saturation and contrast of an image . | {
"entities": {
"method": [
{
"text": "data augmentation",
"start": 42,
"end": 59
},
{
"text": "horizontal flipping",
"start": 88,
"end": 107
},
{
"text": "random scaling",
"start": 110,
"end": 124
},
... | {
"entities": [
{
"label": "dataset",
"description": "A realistic collection of data that is used for training, validating, or testing the algorithms. These datasets can consist of various forms of data such as text, images, videos, or structured data. For example, MNIST, COCO, AGNews, IMDb, etc.",
... |
We first apply non - maximal suppression ( NMS ) by using a 3 × 3 max pooling layer on the corner heatmaps . | {
"entities": {
"method": [
{
"text": "non - maximal suppression",
"start": 15,
"end": 40
},
{
"text": "NMS",
"start": 43,
"end": 46
},
{
"text": "3 × 3 max pooling",
"start": 60,
"end": 77
}
]
},
"... | {
"entities": [
{
"label": "dataset",
"description": "A realistic collection of data that is used for training, validating, or testing the algorithms. These datasets can consist of various forms of data such as text, images, videos, or structured data. For example, MNIST, COCO, AGNews, IMDb, etc.",
... |
We evaluate CornerNet on the very challenging MS COCO dataset ( Lin et al. , 2 0 1 4 ) . | {
"entities": {
"method": [
{
"text": "CornerNet",
"start": 12,
"end": 21
}
],
"dataset": [
{
"text": "MS COCO",
"start": 46,
"end": 53
}
]
},
"relations": {
"evaluated with": [
{
"head": {
"text": ... | {
"entities": [
{
"label": "dataset",
"description": "A realistic collection of data that is used for training, validating, or testing the algorithms. These datasets can consist of various forms of data such as text, images, videos, or structured data. For example, MNIST, COCO, AGNews, IMDb, etc.",
... |
MS COCO uses average precisions ( APs ) at different IoUs and APs for different object sizes as the main evaluation metrics . Corner pooling is a key component of CornerNet . | {
"entities": {
"dataset": [
{
"text": "MS COCO",
"start": 0,
"end": 7
}
],
"method": [
{
"text": "Corner pooling",
"start": 129,
"end": 143
},
{
"text": "CornerNet",
"start": 166,
"end": 175
}
... | {
"entities": [
{
"label": "dataset",
"description": "A realistic collection of data that is used for training, validating, or testing the algorithms. These datasets can consist of various forms of data such as text, images, videos, or structured data. For example, MNIST, COCO, AGNews, IMDb, etc.",
... |
Corner pooling gives similar improvement to corners at different quadrants , show that corner pooling is effective and stable over both small and large areas . | {
"entities": {
"method": [
{
"text": "Corner pooling",
"start": 0,
"end": 14
},
{
"text": "corner pooling",
"start": 87,
"end": 101
}
]
},
"relations": {}
} | {
"entities": [
{
"label": "dataset",
"description": "A realistic collection of data that is used for training, validating, or testing the algorithms. These datasets can consist of various forms of data such as text, images, videos, or structured data. For example, MNIST, COCO, AGNews, IMDb, etc.",
... |
CornerNet uses the hourglass network ( Newell et al. , 2 0 1 6 ) as its backbone network . | {
"entities": {
"method": [
{
"text": "CornerNet",
"start": 0,
"end": 9
},
{
"text": "hourglass network",
"start": 19,
"end": 36
}
]
},
"relations": {
"part of": [
{
"head": {
"text": "hourglass network",
... | {
"entities": [
{
"label": "dataset",
"description": "A realistic collection of data that is used for training, validating, or testing the algorithms. These datasets can consist of various forms of data such as text, images, videos, or structured data. For example, MNIST, COCO, AGNews, IMDb, etc.",
... |
Since the hourglass network is not commonly used in other state - of - the - art detectors , we perform an experiment to study the contribution of the hourglass network in CornerNet . | {
"entities": {
"method": [
{
"text": "hourglass network",
"start": 10,
"end": 27
},
{
"text": "hourglass network",
"start": 151,
"end": 168
},
{
"text": "CornerNet",
"start": 172,
"end": 181
}
]
},
... | {
"entities": [
{
"label": "dataset",
"description": "A realistic collection of data that is used for training, validating, or testing the algorithms. These datasets can consist of various forms of data such as text, images, videos, or structured data. For example, MNIST, COCO, AGNews, IMDb, etc.",
... |
We train a CornerNet in which we replace the hourglass network with FPN ( w/ ResNet - 1 0 1 ) , which is more commonly used in state - of - the - art object detectors . | {
"entities": {
"method": [
{
"text": "CornerNet",
"start": 11,
"end": 20
},
{
"text": "hourglass network",
"start": 45,
"end": 62
},
{
"text": "FPN",
"start": 68,
"end": 71
},
{
"text": "ResN... | {
"entities": [
{
"label": "dataset",
"description": "A realistic collection of data that is used for training, validating, or testing the algorithms. These datasets can consist of various forms of data such as text, images, videos, or structured data. For example, MNIST, COCO, AGNews, IMDb, etc.",
... |
Tab . 4 shows that CornerNet with hourglass network outperforms CornerNet with FPN by 8. 2 % AP , and the anchor box based detector with hourglass network by 5. 5 % AP . | {
"entities": {
"method": [
{
"text": "CornerNet with hourglass network",
"start": 19,
"end": 51
},
{
"text": "CornerNet with FPN",
"start": 64,
"end": 82
},
{
"text": "anchor box based detector with hourglass network",
... | {
"entities": [
{
"label": "dataset",
"description": "A realistic collection of data that is used for training, validating, or testing the algorithms. These datasets can consist of various forms of data such as text, images, videos, or structured data. For example, MNIST, COCO, AGNews, IMDb, etc.",
... |
The results suggest that the choice of the backbone network is important and the hourglass network is crucial to the performance of CornerNet . A good detector should predict high quality bounding boxes that cover objects tightly . | {
"entities": {
"method": [
{
"text": "hourglass network",
"start": 81,
"end": 98
},
{
"text": "CornerNet",
"start": 132,
"end": 141
}
]
},
"relations": {
"part of": [
{
"head": {
"text": "hourglass network... | {
"entities": [
{
"label": "dataset",
"description": "A realistic collection of data that is used for training, validating, or testing the algorithms. These datasets can consist of various forms of data such as text, images, videos, or structured data. For example, MNIST, COCO, AGNews, IMDb, etc.",
... |
To understand the quality of the bounding boxes predicted by CornerNet , we evaluate the performance of CornerNet at multiple IoU thresholds , and compare the results with other state - of - the - art detectors , including RetinaNet , Cascade R - CNN ( Cai and Vasconcelos , 2 0 1 7 ) and IoU - Net ( Jiang et al. , 2 0 ... | {
"entities": {
"method": [
{
"text": "CornerNet",
"start": 61,
"end": 70
},
{
"text": "CornerNet",
"start": 104,
"end": 113
},
{
"text": "RetinaNet",
"start": 223,
"end": 232
},
{
"text": "Ca... | {
"entities": [
{
"label": "dataset",
"description": "A realistic collection of data that is used for training, validating, or testing the algorithms. These datasets can consist of various forms of data such as text, images, videos, or structured data. For example, MNIST, COCO, AGNews, IMDb, etc.",
... |
Tab . 5 shows that CornerNet achieves a much higher AP at 0. 9 IoU than other detectors , outperforming Cascade R - CNN + IoU - Net by 3. 9 % , Cascade R - CNN by 7. 6 % and RetinaNet 2 by 7. 3 % . | {
"entities": {
"method": [
{
"text": "CornerNet",
"start": 19,
"end": 28
},
{
"text": "Cascade R - CNN + IoU - Net",
"start": 104,
"end": 131
},
{
"text": "Cascade R - CNN",
"start": 144,
"end": 159
},
... | {
"entities": [
{
"label": "dataset",
"description": "A realistic collection of data that is used for training, validating, or testing the algorithms. These datasets can consist of various forms of data such as text, images, videos, or structured data. For example, MNIST, COCO, AGNews, IMDb, etc.",
... |
CornerNet simultaneously outputs heatmaps , offsets , and embeddings , all of which affect detection performance . | {
"entities": {
"method": [
{
"text": "CornerNet",
"start": 0,
"end": 9
}
]
},
"relations": {}
} | {
"entities": [
{
"label": "dataset",
"description": "A realistic collection of data that is used for training, validating, or testing the algorithms. These datasets can consist of various forms of data such as text, images, videos, or structured data. For example, MNIST, COCO, AGNews, IMDb, etc.",
... |
We compare CornerNet with other state - of - the - art detectors on MS COCO test - dev ( Tab . 7 ) . | {
"entities": {
"method": [
{
"text": "CornerNet",
"start": 11,
"end": 20
}
],
"dataset": [
{
"text": "MS COCO test - dev",
"start": 68,
"end": 86
}
]
},
"relations": {
"evaluated with": [
{
"head": {
... | {
"entities": [
{
"label": "dataset",
"description": "A realistic collection of data that is used for training, validating, or testing the algorithms. These datasets can consist of various forms of data such as text, images, videos, or structured data. For example, MNIST, COCO, AGNews, IMDb, etc.",
... |
We have presented CornerNet , a new approach to object detection that detects bounding boxes as pairs of corners . | {
"entities": {
"method": [
{
"text": "CornerNet",
"start": 18,
"end": 27
}
],
"task": [
{
"text": "object detection",
"start": 48,
"end": 64
}
]
},
"relations": {
"used for": [
{
"head": {
"text": ... | {
"entities": [
{
"label": "dataset",
"description": "A realistic collection of data that is used for training, validating, or testing the algorithms. These datasets can consist of various forms of data such as text, images, videos, or structured data. For example, MNIST, COCO, AGNews, IMDb, etc.",
... |
We evaluate CornerNet on MS COCO and demonstrate competitive results . | {
"entities": {
"method": [
{
"text": "CornerNet",
"start": 12,
"end": 21
}
],
"dataset": [
{
"text": "MS COCO",
"start": 25,
"end": 32
}
]
},
"relations": {
"evaluated with": [
{
"head": {
"text": ... | {
"entities": [
{
"label": "dataset",
"description": "A realistic collection of data that is used for training, validating, or testing the algorithms. These datasets can consist of various forms of data such as text, images, videos, or structured data. For example, MNIST, COCO, AGNews, IMDb, etc.",
... |
High - resolution representation learning plays an essential role in many vision problems , e.g. , pose estimation and semantic segmentation . | {
"entities": {
"method": [
{
"text": "High - resolution representation learning",
"start": 0,
"end": 41
}
],
"task": [
{
"text": "pose estimation",
"start": 99,
"end": 114
},
{
"text": "semantic segmentation",
"... | {
"entities": [
{
"label": "dataset",
"description": "A realistic collection of data that is used for training, validating, or testing the algorithms. These datasets can consist of various forms of data such as text, images, videos, or structured data. For example, MNIST, COCO, AGNews, IMDb, etc.",
... |
The high - resolution network (HRNet)~\cite{SunXLW 1 9 } , recently developed for human pose estimation , maintains high - resolution representations through the whole process by connecting high - to - low resolution convolutions in \emph{parallel } and produces strong high - resolution representations by repeatedly co... | {
"entities": {
"method": [
{
"text": "high - resolution network",
"start": 4,
"end": 29
},
{
"text": "(HRNet)~\\cite{SunXLW",
"start": 30,
"end": 50
},
{
"text": "convolutions",
"start": 217,
"end": 229
},... | {
"entities": [
{
"label": "dataset",
"description": "A realistic collection of data that is used for training, validating, or testing the algorithms. These datasets can consist of various forms of data such as text, images, videos, or structured data. For example, MNIST, COCO, AGNews, IMDb, etc.",
... |
We show top results in semantic segmentation on Cityscapes , LIP , and PASCAL Context , and facial landmark detection on AFLW , COFW , $ 3 0 0 $W , and WFLW . | {
"entities": {
"task": [
{
"text": "semantic segmentation",
"start": 23,
"end": 44
},
{
"text": "facial landmark detection",
"start": 92,
"end": 117
}
],
"dataset": [
{
"text": "Cityscapes",
"start": 48,
... | {
"entities": [
{
"label": "dataset",
"description": "A realistic collection of data that is used for training, validating, or testing the algorithms. These datasets can consist of various forms of data such as text, images, videos, or structured data. For example, MNIST, COCO, AGNews, IMDb, etc.",
... |
In addition , we build a multi - level representation from the high - resolution representation and apply it to the Faster R - CNN object detection framework and the extended frameworks . | {
"entities": {
"method": [
{
"text": "Faster R - CNN",
"start": 116,
"end": 130
}
],
"task": [
{
"text": "object detection",
"start": 131,
"end": 147
}
]
},
"relations": {
"used for": [
{
"head": {
... | {
"entities": [
{
"label": "dataset",
"description": "A realistic collection of data that is used for training, validating, or testing the algorithms. These datasets can consist of various forms of data such as text, images, videos, or structured data. For example, MNIST, COCO, AGNews, IMDb, etc.",
... |
The proposed approach achieves superior results to existing single - model networks on COCO object detection . | {
"entities": {
"dataset": [
{
"text": "COCO",
"start": 87,
"end": 91
}
],
"task": [
{
"text": "object detection",
"start": 92,
"end": 108
}
]
},
"relations": {
"benchmark for": [
{
"head": {
"text"... | {
"entities": [
{
"label": "dataset",
"description": "A realistic collection of data that is used for training, validating, or testing the algorithms. These datasets can consist of various forms of data such as text, images, videos, or structured data. For example, MNIST, COCO, AGNews, IMDb, etc.",
... |
There are two main kinds of representations : low - resolution representations that are mainly for image classification , and high - resolution representations that are essential for many other vision problems , e.g. , semantic segmentation , object detection , human pose estimation , etc . | {
"entities": {
"task": [
{
"text": "image classification",
"start": 99,
"end": 119
},
{
"text": "semantic segmentation",
"start": 219,
"end": 240
},
{
"text": "object detection",
"start": 243,
"end": 259
}... | {
"entities": [
{
"label": "dataset",
"description": "A realistic collection of data that is used for training, validating, or testing the algorithms. These datasets can consist of various forms of data such as text, images, videos, or structured data. For example, MNIST, COCO, AGNews, IMDb, etc.",
... |
One is to recover high - resolution representations from low - resolution representations outputted by a network ( e.g. , ResNet ) and optionally intermediate medium - resolution representations , e.g. , Hourglass [ 7 2 ] , SegNet [ 2 ] , DeconvNet [ 7 4 ] , U - Net [ 8 3 ] , and encoderdecoder [ 7 7 ] . | {
"entities": {
"method": [
{
"text": "ResNet",
"start": 122,
"end": 128
},
{
"text": "Hourglass",
"start": 204,
"end": 213
},
{
"text": "SegNet",
"start": 224,
"end": 230
},
{
"text": "Deconv... | {
"entities": [
{
"label": "dataset",
"description": "A realistic collection of data that is used for training, validating, or testing the algorithms. These datasets can consist of various forms of data such as text, images, videos, or structured data. For example, MNIST, COCO, AGNews, IMDb, etc.",
... |
In addition , dilated convolutions are used to replace some strided convolutions and associated regular convolutions in classification networks to compute medium - resolution representations [ 1 3 , 1 2 6 ] . | {
"entities": {
"method": [
{
"text": "dilated convolutions",
"start": 14,
"end": 34
},
{
"text": "strided convolutions",
"start": 60,
"end": 80
},
{
"text": "convolutions",
"start": 104,
"end": 116
}
]... | {
"entities": [
{
"label": "dataset",
"description": "A realistic collection of data that is used for training, validating, or testing the algorithms. These datasets can consist of various forms of data such as text, images, videos, or structured data. For example, MNIST, COCO, AGNews, IMDb, etc.",
... |
End of preview. Expand in Data Studio
SciER
SciER (named entity recognition and relation extraction), rebuilt for MNEB from the original release with character-level offsets.
Source: https://github.com/edzq/SciER (commit db347813, SciER/PLM/*.jsonl)
| config | split | records |
|---|---|---|
| in_domain | train | 5575 |
| in_domain | validation | 713 |
| in_domain | test | 854 |
| ood | test | 580 |
Labels
entities: dataset, method, task
relations: part of, subclass of, subtask of, benchmark for, trained with, evaluated with, synonym of, used for, compare with
Format
Each record has input (the text), output (gold annotations grouped by label) and schema (the label set with descriptions). Offsets are character offsets into input, half-open [start, end).
"output": {
"entities": {"<label>": [{"text", "start", "end"}]},
"relations": {"<label>": [{"head": {"text", "start", "end"},
"tail": {"text", "start", "end"}}]},
"events": {"<label>": [{"trigger": {"text", "start", "end"},
"arguments": [{"role", "text", "start", "end"}]}]}
}
Each schema item has a description and a description_source: official when adapted from the dataset's paper or guideline, mneb when written for MNEB. Relation datasets also give setting (end_to_end or gold_entity) and negative_label; relation items give symmetric. Event items list the roles valid for that event type.
Within a record, identical annotations are kept once and outer whitespace is trimmed from spans; source annotations are otherwise kept as annotated.
Notes
- Configs
in_domain(train/validation/test) andood(test). Sentence level. Not SciERC. - Descriptions are adapted from the SciER annotation guideline.
Build
Built by scripts/datasets/sci_er.py in the MNEB repository.
Citation
Zhang et al. (2024), SciER: An Entity and Relation Extraction Dataset for Datasets, Methods, and Tasks in Scientific Documents. EMNLP.
License: gpl-3.0
- Downloads last month
- -