AI Inference Workspace

Analyze a Chili Photo

Detect every chili in a single image and classify plant health in seconds. Powered by a two-stage YOLOv11 + EfficientNetV2-S pipeline.

Real-time detection

Instance segmentation with YOLOv11

Health classifier

EfficientNetV2-S binary head

Saved reports

Exportable PDF, full history

Detection accuracy

92%+

on validation set

Avg inference

~90s

per image (CPU hosted)

Pipeline

2-stage

detect + classify

Upload
Analyze
Results

Upload an image

Drop a photo or browse your files

Step 1 of 3

Drag & drop or click to browse

PNG, JPG or phone photo · auto-optimized for upload

Upload or pick a sample to enable analysis.

Don't have an image?

Tips for accurate results

  • Use natural, even lighting — avoid harsh shadows.
  • Frame the plant or fruit clearly in the centre.
  • Use 1024px+ photos for best detection accuracy.

Under the hood

How ChiliSense analyzes your photo

A purpose-built two-stage pipeline for fast, explainable results.

1

Stage 1

Detect

YOLOv11-seg

Instance segmentation finds every chili pepper in the frame and outputs a precise mask for each one.

2

Stage 2

Classify

EfficientNetV2-S

Each detected chili is cropped and passed through a binary classifier that flags healthy vs unhealthy.

3

Stage 3

Report

Saved + exportable

Results are stored in your account with per-instance breakdowns and an exportable PDF report.