Research conferences, 20 to 25 Jun 2027
CVPR
The Conference on Computer Vision and Pattern Recognition, where much vision-language model research appears.
CVPR, first held in 1983, is the largest computer vision conference. As models have learned to read images alongside text, it has become a main venue for vision-language research: how models describe images, answer questions about photos and documents, follow video and act in the world. Image encoders used inside multimodal language models are often first described there. CVPR 2027 is in Seattle, with workshops and tutorials on 20 and 21 June and the main conference from 22 to 25 June.
Models tied to CVPR
Still to be written: GPT-4o.