CORE TECHNOLOGY

Proprietary Vision AI Engine:
BV-Pose

High-performance Pose Estimation technology that precisely captures
human movements even in extreme industrial environments.

import biovis_ai as bv
model = bv.PoseEstimator(mode="industrial_top_view")
keypoints = model.inference(frame) # Returns 15 keypoints with high confidence

Core Competencies

Optimized for Top-View

Precisely tracks 15 worker keypoints without error, even in high-angle and wide-angle distortion environments typical of factory CCTVs, beyond standard frontal angles.

Adaptive Performance

Provides flexibility regardless of on-site hardware specs. We possess a 3-stage optimization pipeline ranging from CPU-exclusive Nano models to GPU-accelerated high-precision Large models.

Robust Occlusion Handling

Applies a proprietary tracking algorithm that minimizes ID Switching and infers occluded body parts, even in complex facilities mixed with multiple workers.

Technical Specifications

Architecture
Custom YOLOv8-based Network
Input Resolution
416px ~ 640px (Dynamic Scaling)
Inference Speed
Max 120 FPS (on NVIDIA GPU)
Output Format
Normalized Keypoint Coordinates (JSON/NPY)