CJ's Video Understanding AI Tech Accepted at Top Deep Learning Conference

CJ Group's artificial intelligence (AI) technology has been recognized at the world's most authoritative deep learning conference. CJ Group announced that a paper related to AI technology that identifies story flow within videos and automatically segments scenes has been finally accepted by the 14th International Conference on Learning Representations (ICLR 2026), scheduled for 2026. ICLR is a prestigious conference researching how AI learns data, and this acceptance demonstrates CJ's competitiveness in video understanding AI technology.

This technology has advanced to the level of understanding the meaning and story of videos, and can accurately segment scenes by analyzing genre information such as comedy or thriller, as well as the editing length of the video.

CJ Group is applying this technology across its content business to enhance production efficiency. CJ 4DPLEX utilizes it to improve the precision and efficiency of 4DX and ScreenX (SX) production processes through scene-by-scene analysis. Studio Dragon enhances digital asset management (DAM) efficiency by automatically classifying and analyzing its vast video library on a scene-by-scene basis. CJ ONSTYLE integrates this technology into mobile live commerce content production, automatically extracting and processing key promotional segments for each product from long original videos. The AI·DT Promotion Office has also newly built MovieChat-SSeg, an evolution of the existing video understanding dataset MovieChat.