AJ Robotics

Checking...
Dashboard Machines
Robotics ▾
Overview Gamepad Test Vision Training
Trader Deploy Training
Datasets
Training
Models
Object Detection
Unsupervised

Datasets

Name Type Classes Train Val Modified
Loading...

Training Setup

Dreamer GPU Status

GPU--
Utilization--
Memory--
Training Service--

Trained Models

Loading...

R1 Camera Capture

Threshold: 0.25

Save for Training

Capture images and save to a dataset folder for labeling

Robot Status

R1Checking...

Detection Result

📷
Click "Capture & Detect" to capture from R1 and run YOLO detection

Detection History

No detections yet

Unsupervised Learning

No labels needed — discover patterns, cluster images, or detect anomalies.

About Unsupervised Learning

Image Clustering
Groups similar images together without labels. Uses YOLO backbone to extract visual features, then K-Means to find clusters. Useful for organizing datasets, discovering object categories, or finding duplicate/similar images.
Anomaly Detection
Trains an autoencoder on "normal" images. Images that can't be reconstructed well are flagged as anomalies. Useful for quality control, defect detection, or finding unusual objects in a scene.
Similar Image Search
Upload or capture a query image, and find the most similar images in a dataset. Uses cosine similarity on YOLO feature vectors.
AJ Robotics © 2026 | YOLO Training on Dreamer (RTX 4070)