Skip to content
Agientry
ZH
← Back to Videos

A 1.9B Vision Model Watches Café Tables

Loading video
Loading video

A developer built a local café-monitoring app around the 1.9B d1-2B model from Liquid AI. It tracks free, occupied and dirty tables from one camera, runs on a laptop without cloud inference, and generates no output tokens.

Vision-language modelsEdge AIComputer VisionRestaurant OperationsLiquid AILFM2-VL
Category: perception
Author: @romankhrupa
Date: 2026-11-09T00:00:00
Duration: 32.816s
Source: https://x.com/romankhrupa/status/2119913625511137746

Related Videos

Sixteen Live Camera Streams on One Edge System
0:19

Sixteen Live Camera Streams on One Edge System

@ultralytics · 25 views · 2026-11-09
Computer VisionMulti-Camera
Reka EdgeQ: An On-Device VLM Running Natively on the Snapdragon Hexagon NPU
0:07

Reka EdgeQ: An On-Device VLM Running Natively on the Snapdragon Hexagon NPU

@RekaAILabs · 736 views · 2026-09-22
VLMOn-Device Inference
NanoOWL: Zero-Shot Real-Time Object Detection by NVIDIA
0:14

NanoOWL: Zero-Shot Real-Time Object Detection by NVIDIA

@simplifyinAI · 597 views · 2026-08-24
NVIDIANanoOWL
InstantHMR Brings Live Human Mesh Recovery to ComfyUI
1:26

InstantHMR Brings Live Human Mesh Recovery to ComfyUI

@jtydhr88 · 92 views · 2026-11-07
Human Mesh RecoveryReal-Time Perception
ABot-Recon Streams 3D Reconstruction over Long Video
0:27

ABot-Recon Streams 3D Reconstruction over Long Video

@cutetoxicguy · 77 views · 2026-11-07
3D reconstructionSLAM
EDGAR Refines Gaussian Head Avatars with Identity-Guided Diffusion
0:10

EDGAR Refines Gaussian Head Avatars with Identity-Guided Diffusion

@Machinedelusion · 65 views · 2026-11-07
Gaussian SplattingHead Avatar

As an Amazon Associate, we earn from qualifying purchases.