APLA Inc.

에이플라; 주식회사 에이플라

"APLA Inc. is a South Korean computer vision software company founded in October 2024 and based at the Pangyo Techno Valley Startup Campus in Seongnam, Gyeonggi Province. It is led by chief executive Lee Yong-soo and had six employees as of July 2026. The company builds a generative three-dimensional motion engine that extracts full-body human motion from ordinary two-dimensional video, and it sells that engine into two markets: robot motion learning and digital avatar content.

The core technical claim is that high-quality 3D motion data can be produced from software alone, without LiDAR, depth sensors or marker suits. The engine runs in real time on standard smartphone cameras, webcams and low-specification PCs across iOS and Android, and the company states that inference is performed on-device rather than on a server. APLA reports a PA-MPJPE score of 22.5 on the 3DPW benchmark, a standard public measure of 3D human pose estimation error, and says the model is trained on a labelled motion dataset it is building toward more than ten million entries. The company holds nine granted Korean patents and one granted United States patent in computer vision, with further applications pending in Japan and the United States.

The robotics product converts human movement captured from 2D video into robot joint control sequences, so that a demonstration performed in front of a camera can be turned into machine-readable motion for a robot to reproduce. The company names humanoids, collaborative robots and service robots as the target platforms, and positions the output as training data for robot learning as well as direct control input. This places APLA in the demonstration-to-action layer of embodied AI rather than in robot hardware: it supplies the motion data and retargeting step that sits between a human demonstration and a robot controller.

The content side of the business consists of three products. APLA Studio is a desktop application for creating and broadcasting virtual avatars. APLA Mobile is a smartphone tool for short-form content that performs full-body tracking without external hardware, which the company describes as a first for mobile full-tracking live streaming. A generative avatar feature produces a talking 3D avatar with facial expression and body motion from a single user image plus voice data, aimed at chatbot and virtual human deployments. Customers on this side are described as multi-channel networks and individual creators; none are named publicly.

APLA raised a seed round from Seoul National University Technology Holdings announced on 17 March 2025; the amount was not disclosed. The company is a member of the NVIDIA Inception program. It has stated plans to establish a Japanese subsidiary, raise a Series A and exhibit at CES as part of a North American expansion, though none of these had been confirmed as completed at the time of collection.

The company positions itself as an infrastructure supplier to the AI-generated 3D content and robotics industries rather than as an end-product vendor, and its stated differentiator against conventional motion capture is cost and accessibility: no capture studio, no depth hardware and no marker suit."