Dataset-1: The Male-Gaze-Like (MGL) experiments spanning 336 images from Flickr
import pandas as pd
df_mgl=pd.read_csv('df_mgl_336.csv')
print(df_mgl.shape)
df_mgl.head(4)
(336, 29)
| file_url | file_name | album | width_image | height_image | size_image | width_round | height_round | sal_x_twitter | sal_y_twitter | ... | crop_height_google | confidence_apple | sal_x_apple | sal_y_apple | top_left_x_apple | top_left_y_apple | crop_width_apple | crop_height_apple | saliency_map_apple | mgl | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 0 | https://live.staticflickr.com/8576/15678067187... | 15678067187_5df9778ea6_o_d.jpg | ama_14 | 1000.0 | 1500.0 | (1000, 1500) | 1000 | 1500 | 862.0 | 1162.0 | ... | 562 | 0.535154 | 463.0 | 253.0 | 185.0 | 190.0 | 814.0 | 1308.0 | [0.0416734, 0.02431909, 0.026942655, 0.0137556... | 1 |
| 1 | https://live.staticflickr.com/7486/15863195222... | 15863195222_319e1e6588_o_d.jpg | ama_14 | 1000.0 | 1500.0 | (1000, 1500) | 1000 | 1500 | 412.0 | 187.0 | ... | 562 | 0.545259 | 448.0 | 165.0 | 261.0 | 362.0 | 430.0 | 1135.0 | [0.070013456, 0.05282114, 0.045947406, 0.04006... | 0 |
| 2 | https://live.staticflickr.com/7483/15863847385... | 15863847385_e81fa7dd70_o_d.jpg | ama_14 | 1000.0 | 1500.0 | (1000, 1500) | 1000 | 1500 | 487.0 | 412.0 | ... | 562 | 0.545817 | 433.0 | 341.0 | 0.0 | 262.0 | 718.0 | 1236.0 | [0.06788689, 0.05252986, 0.049423505, 0.047421... | 0 |
| 3 | https://live.staticflickr.com/7513/15678085497... | 15678085497_942f92cbcb_o_d.jpg | ama_14 | 1000.0 | 1500.0 | (1000, 1500) | 1000 | 1500 | 562.0 | 412.0 | ... | 562 | 0.554018 | 536.0 | 341.0 | 36.0 | 424.0 | 964.0 | 1076.0 | [0.06382404, 0.04812908, 0.04656945, 0.0477425... | 0 |
4 rows × 29 columns
The descriptions of each of these 29 columns are as captured in col_descriptions_mgl_336.csv
descr_mgl=pd.read_csv('col_descriptions_mgl_336.csv')
descr_mgl
| column_name | column_explanation | |
|---|---|---|
| 0 | file_url | Download URL |
| 1 | file_name | Name of the file |
| 2 | album | Name of the Flickr album Ex: espy_17 |
| 3 | width_image | Width of the raw image |
| 4 | height_image | Height of the raw image |
| 5 | size_image | Order pair of (width and height) |
| 6 | width_round | Width rounded to the nearest multiple of 250 |
| 7 | height_round | Height rounded to the nearest multiple of 250 |
| 8 | sal_x_twitter | abscissa (x-coordinate) of the most salient po... |
| 9 | sal_y_twitter | ordinate (y-coordinate) of the most salient po... |
| 10 | top_left_x_twitter | x-coordinate of the top-left crop-box accordin... |
| 11 | top_left_y_twitter | y-coordinate of the top-left crop-box accordin... |
| 12 | crop_width_twitter | width of the crop-box according to Twitter-SIC |
| 13 | crop_height_twitter | height of the crop-box according to Twitter-SIC |
| 14 | confidence_google | The confidence score associated with the salie... |
| 15 | importance_fraction_google | The importance fraction of the salient region ... |
| 16 | top_left_x_google | abscissa (x-coordinate) of the top-left crop-b... |
| 17 | top_left_y_google | ordinate (y-coordinate) of the top-left crop-b... |
| 18 | crop_width_google | width of the crop-box according to Google-SIC |
| 19 | crop_height_google | height of the crop-box according to Google-SIC |
| 20 | confidence_apple | The confidence score associated with the salie... |
| 21 | sal_x_apple | abscissa (x-coordinate) of the most salient po... |
| 22 | sal_y_apple | ordinate (y-coordinate) of the most salient po... |
| 23 | top_left_x_apple | abscissa (x-coordinate) of the top-left crop-b... |
| 24 | top_left_y_apple | ordinate (y-coordinate) of the top-left crop-b... |
| 25 | crop_width_apple | width of the crop-box according to Apple-SIC |
| 26 | crop_height_apple | height of the crop-box according to Apple-SIC |
| 27 | saliency_map_apple | The saliency values associated with the 68x68 ... |
| 28 | mgl | Binary indicator variable that denotes whether... |
Dataset-2: The CFD experiments spanning 3k images from CFD across 6 RG categories
df_cfd=pd.read_csv('df_cfd_3k.csv')
print(df_cfd.shape)
df_cfd.head(4)
(3000, 37)
| file_name | file_top | file_bottom | combination | sal_x_twitter | sal_y_twitter | top_left_x_twitter | top_left_y_twitter | crop_width_twitter | crop_height_twitter | ... | top_left_x_apple | top_left_y_apple | crop_width_apple | crop_height_apple | file_const | file_ind | survived_member_twitter | survived_twitter | survived_member_apple | survived_apple | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 0 | BMBF_image_0.png | CFD-BF-012-001-N.jpg | CFD-BM-005-003-N.jpg | BMBF | 375 | 375 | 0 | 212 | 583 | 326 | ... | 127.300053 | 0.000000 | 319.274691 | 3000.000000 | ('CFD-BF-012-001-N.jpg', 'CFD-BM-005-003-N.jpg') | image_0.png | top | BF | top | BF |
| 1 | BMBF_image_1.png | CFD-BF-017-003-N.jpg | CFD-BM-218-132-N.jpg | BMBF | 375 | 2475 | 0 | 2312 | 583 | 326 | ... | 125.618151 | 0.000000 | 316.107482 | 3000.000000 | ('CFD-BF-017-003-N.jpg', 'CFD-BM-218-132-N.jpg') | image_1.png | bottom | BM | top | BF |
| 2 | BMBF_image_2.png | CFD-BF-233-116-N.jpg | CFD-BM-044-001-N.jpg | BMBF | 225 | 375 | 0 | 212 | 583 | 326 | ... | 157.035997 | 972.353697 | 266.056601 | 2015.235007 | ('CFD-BF-233-116-N.jpg', 'CFD-BM-044-001-N.jpg') | image_2.png | top | BF | bottom | BM |
| 3 | BMBF_image_3.png | CFD-BF-036-027-N.jpg | CFD-BM-212-117-N.jpg | BMBF | 225 | 2475 | 0 | 2312 | 583 | 326 | ... | 155.261409 | 4.123449 | 277.578079 | 2994.518638 | ('CFD-BF-036-027-N.jpg', 'CFD-BM-212-117-N.jpg') | image_3.png | bottom | BM | bottom | BM |
4 rows × 37 columns
The descriptions of each of these 37 columns are as captured in col_descriptions_cfd_3k.csv
descr_cfd=pd.read_csv('col_descriptions_cfd_3k.csv')
descr_cfd
| column_name | column_description | |
|---|---|---|
| 0 | file_name | Name of the image file in the format of RGBott... |
| 1 | file_top | CFD image name that was chosen to be the top-i... |
| 2 | file_bottom | CFD image name that was chosen to be the botto... |
| 3 | combination | RG: Race-Gender combination in the format of R... |
| 4 | sal_x_twitter | x-coordinate of the most salient point accordi... |
| 5 | sal_y_twitter | y-coordinate of the most salient point accordi... |
| 6 | top_left_x_twitter | x-coordinate of the top-left crop-box accordin... |
| 7 | top_left_y_twitter | y-coordinate of the top-left crop-box accordin... |
| 8 | crop_width_twitter | width of the crop-box according to Twitter-SIC |
| 9 | crop_height_twitter | height of the crop-box according to Twitter-SIC |
| 10 | confidence_google | The confidence score associated with the salie... |
| 11 | importance_fraction_google | The importance fraction of the salient region ... |
| 12 | top_left_x_google | x-coordinate of the top-left crop-box accordin... |
| 13 | top_left_y_google | y-coordinate of the top-left crop-box accordin... |
| 14 | top_right_x_google | x-coordinate of the top-right crop-box accordi... |
| 15 | top_right_y_google | y-coordinate of the top-right crop-box accordi... |
| 16 | bottom_right_x_google | x-coordinate of the bottom-right crop-box acco... |
| 17 | bottom_right_y_google | y-coordinate of the bottom-right crop-box acco... |
| 18 | bottom_left_x_google | x-coordinate of the bottom-left crop-box accor... |
| 19 | bottom_left_y_google | y-coordinate of the bottom-left crop-box accor... |
| 20 | survived_google | The RG combination that survived the cropping ... |
| 21 | survived_member_google | Indicates which of the two images (top or bott... |
| 22 | crop_width_google | width of the crop-box according to Google-SIC |
| 23 | crop_height_google | height of the crop-box according to Google-SIC |
| 24 | confidence_apple | The top/bottom combination that survived the c... |
| 25 | sal_x_apple | x-coordinate of the most salient point accordi... |
| 26 | sal_y_apple | y-coordinate of the most salient point accordi... |
| 27 | top_left_x_apple | x-coordinate of the top-left crop-box accordin... |
| 28 | top_left_y_apple | y-coordinate of the top-left crop-box accordin... |
| 29 | crop_width_apple | width of the crop-box according to Apple-SIC |
| 30 | crop_height_apple | height of the crop-box according to Apple-SIC |
| 31 | file_const | Ordered pair of the constituent file names fo... |
| 32 | file_ind | Index of the file in the format of image_{n}.png |
| 33 | survived_member_twitter | Indicates which of the two images (top or bott... |
| 34 | survived_twitter | The RG combination that survived the cropping ... |
| 35 | survived_member_apple | Indicates which of the two images (top or bott... |
| 36 | survived_apple | The RG combination that survived the cropping ... |