TAPe+ML — Efficient Computer Vision
Collection
Compact Computer Vision for detection, segmentation, classification and Edge AI. • 3 items • Updated
TAPe+ML
Efficient Computer Vision for Detection, Segmentation and Classification
TAPe+ML is a proprietary Computer Vision architecture developed by Comexp Research Lab.
It provides object detection, image segmentation and image classification in a compact architecture designed for efficient CPU, Edge and on-premise deployment.
Key Features
- Detection, segmentation and classification in one architecture
- Less than 100K parameters
- Model weight below 100 KB
- Approximately 8 ms per frame on CPU
- CPU and Edge deployment
- Training and adaptation with tens of samples
COCO Benchmark
- mAP50: 84.7%
- mAP50-95: 65.3%
- Segmentation AP50: 80.7%
Explore TAPe+ML
Live Demo
Try TAPe+ML on your own data:
TAPe+ML Platform
Test the platform and use TAPe+ML for your Computer Vision tasks:
Plans & Pricing
See available plans and access options:
User Guide
Learn how to use Detection, Segmentation, Classification, Custom Models, Fine-tuning and the REST API:
Supported Tasks
API Access
TAPe+ML inference and product access are provided through the Comexp platform.
Model weights are proprietary and are not publicly distributed.
For platform access and API integration:
Deployment
TAPe+ML is designed for efficient inference on CPU and Edge devices without requiring a GPU.
Research
TAPe+ML: A Compact Structured Representation for Multi-Task Computer Vision
About Comexp Research Lab
Comexp Research Lab develops efficient Computer Vision and AI technologies for Edge, on-premise and cloud deployments.
Website: https://comexp.net/
TAPe+ML Platform: https://ml.comexp.net/