Tech & Startup

DU hosts AI art exhibition

AI Art Exhibition
The exhibition is an attempt to create awareness about environment pollution through AI image generation. Image: Md. Zahidur Rabbi.

An artificial intelligence (AI) art exhibition is currently being held at the University of Dhaka's Fine Arts Faculty, till June 6. Organised by University of Dhaka's Data and Design Lab, this amalgamation of art and technology is showcasing various images created using Artificial Intelligence (AI).

One of the highlights of this exhibition is 'Demo Passport' where AI creates a 'passport' using a person's photo and information. Firstly, the user's picture, name and some relevant information is taken. The passport is then created using this information as input to the AI. The passport shows an abstract image instead of the user's photo where AI uses the 2023 divisional data of water, air and noise pollution to generate this image. The exhibition is an attempt to create awareness about environment pollution through AI image generation.

According to a press release, the computational time for generating the AI-images involves training the custom-made stable diffusion Image to Image model using the LAION-5B dataset using multiple A100 GPUs for approximately 3 to 4 weeks. The model takes around 1 minute to generate the image on an L4 GPU.

Comments

DU hosts AI art exhibition

AI Art Exhibition
The exhibition is an attempt to create awareness about environment pollution through AI image generation. Image: Md. Zahidur Rabbi.

An artificial intelligence (AI) art exhibition is currently being held at the University of Dhaka's Fine Arts Faculty, till June 6. Organised by University of Dhaka's Data and Design Lab, this amalgamation of art and technology is showcasing various images created using Artificial Intelligence (AI).

One of the highlights of this exhibition is 'Demo Passport' where AI creates a 'passport' using a person's photo and information. Firstly, the user's picture, name and some relevant information is taken. The passport is then created using this information as input to the AI. The passport shows an abstract image instead of the user's photo where AI uses the 2023 divisional data of water, air and noise pollution to generate this image. The exhibition is an attempt to create awareness about environment pollution through AI image generation.

According to a press release, the computational time for generating the AI-images involves training the custom-made stable diffusion Image to Image model using the LAION-5B dataset using multiple A100 GPUs for approximately 3 to 4 weeks. The model takes around 1 minute to generate the image on an L4 GPU.

Comments

বাংলাদেশে গুমের ঘটনায় ভারতের সম্পৃক্ততা খুঁজে পেয়েছে কমিশন

কমিশন জানিয়েছে, আইনশৃঙ্খলা রক্ষাকারী বাহিনীর মধ্যে এ বিষয়ে একটি জোরালো ইঙ্গিত রয়েছে যে, কিছু বন্দি এখনো ভারতের জেলে থাকতে পারে।

৩ ঘণ্টা আগে