| Model Type | Depth Estimation |
|---|---|
| Framework | LiteRT (TensorFlow Lite) |
| Algorithm | MiDaS v2 |
| Technical Details | This unit estimates a dense depth map from a single RGB image using MiDaS v2, optimized via quantization for edge deployment. Outputs are displayed as heatmaps for scene perception and robotic navigation. Based on research by Intel. |
MiDaS for Monocular Depth Estimation
Efficient single-camera depth estimation using quantized MiDaS v2 model optimized for edge AI deployment.
Leave your message here
Fill the form with the required information
$0.00
MiDaS for Monocular Depth Estimation
app details
Resources
Use Cases
Predictive Maintenance
Anomaly Detection
Supported SECO Devices
AI optimized hardware
Fanless embedded PCs
Modules
supported chipsets
Intel® Core™ i3
Qualcomm® Dragonwing QCS6490
Qualcomm® Dragonwing QCS5430
NXP i.MX 8M Plus
NXP i.MX 95
Performance Benchmark
Hardware
Latency (ms)
Memory Usage
Processing unit
SECO SOM-SMARC-QCS6490
123.95 ms
14.77 MB
CPU
SECO SOM-SMARC-QCS6490
8.45 ms
14.77 MB
NPU
SECO SOM SMARC iMX8Plus
1596 ms
14.77 MB
CPU
SECO SOM SMARC iMX8Plus
281.55 ms
14.77 MB
NPU
SECO SOM-SMARC-MX95
356.01 ms
14.77 MB
CPU
SECO Titan 300 TGL-UP3 AI
44.58 ms
14.77 MB
CPU
FFNet for Real-Time Semantic Segmentation
Lightweight semantic segmentation model optimized for real-time street scene analysis on edge devices.
View Model
PoseNet for Multi-Person Pose Detection
Lightweight and quantized PoseNet model for real-time multi-person pose estimation on edge hardware.
View Model
HRNet for Human Pose Estimation
High-resolution neural network for accurate keypoint detection in human pose estimation tasks.
View Model
Time Series Forecast for Electricity Consumption
Neural network model for forecasting household electricity consumption using historical time series data.
View Model|
|