I\u2019m fairly new to reddit so please forgive me if there\u2019s a subreddit this thread would be more suited to!<\/p>\n
Context: I\u2019m currently working on my research proposal paper for a PhD in Fine Arts. I\u2019m primarily a painter, so this is a practice-led research project on the subject of post-photography\/image theory, post-digital visual culture and traumatic representation. I am by no means a data scientist and have a very base level understanding of ML and image recognition, but as I\u2019m exploring traumatic representation in images on the internet\/in relation to screen culture, my work does somewhat intersect with the field of computer vision – which is, of course, what brings me to Reddit. <\/p>\n
I\u2019m interested in how image recognition is used for the automated moderation\/censorship\/removal of \u201csensitive\u201d content on social media platforms. I\u2019m trying to locate any known dataset that\u2019s been used to train this kind of image recognition model – I know there are plenty of datasets specifically for training ML to identify porn, but as my research revolves around trauma I\u2019d ideally like to find one that includes a broader range of NSFW categories (violence, gore, etc.). I\u2019m not too hopeful that any image based dataset of this kind would be publicly accessible (I suppose you\u2019d hope it wasn\u2019t), but alas, just putting this out here if anyone has any leads. <\/p>\n
Even if you can\u2019t answer my question, any thoughts\/feedback\/comments on this are more than welcome. I don\u2019t particularly speak the language of computer science, but always open to having conversations about the project \ud83d\ude42 <\/p>\n<\/div>\n
submitted by \/u\/sentient-glue <\/a>
[link]<\/a><\/span> [comments]<\/a><\/span><\/p>
