The goal of the repository is to host all the code and community data assets we have curated during the execution of our project titled Large datasets: A Pyrrhic win for computer vision?
-
Tour of all the census CSV file generated Tour_csv (nbviewer)
-
InsightFace generate: ImageNet_1A_Insightface_generate (nbviewer)
-
InsightFace analysis: ImageNet_1B_Insightface_analyze (nbviewer)
-
Dex Audit generate: ImageNet_1C_census_audit_dex (nbviewer)
-
NSFW analysis: ImageNet_2_NSFW_analysis (nbviewer)
-
Co-occurence bias analysis (Dog ownership and musical instruments): ImageNet_3_Gender_Dog_musical_instruments_final (nbviewer)
-
Glove embeddings generated: ImageNet_4_Semantics_glove_gen (nbviewer)
-
Semanticity error NSFEW analysis: ImageNet_5C_Semanticity_error (nbviewer)
-
ResnET 50 Accuracies: ImageNet_6A_acc_ResNet50 (nbviewer)
-
NasNet accuracies: ImageNet_6B_acc_NasNet (nbviewer)
-
Coalescing all the results to form a monolithic census dataframe: ImageNet_7_Census_munge (nbviewer)
-
Hand survey images (61 images in the ImageNet dataset): ImageNet_8_Process_hand_survey_files (nbviewer)
-
Tiny Images index analysis : tiny_images_1_index (nbviewer)
-
Tiny images visualizations ( Racial epithets presence warning!) : tiny_images_2_plots (nbviewer)