← Berry Tech
Berry Tech Independent builder's lab
Prithvi-ALS · Prototype / Active
CV Pipeline ↓
Berry Tech Ecosystem
Status: Prototype / Active Development

AI-driven vision for
smarter litter detection.

Prithvi-ALS is an environmental monitoring project exploring how computer vision and edge AI can identify waste in public spaces, school campuses, and parks — enabling proactive sanitation and cleaner public surroundings.

OpenCV & PyTorchComputer Vision
Laptop & EdgeInference Pipeline
Campus & ParksTarget Arenas
PrototypeCurrent Maturity
pipeline — prithvi.cv.detect
v0.3-proto
stream: camera_source:0 (720p @ 24fps) ✓ Frame preprocess: resize, normalize tensor ✓ Inference: Waste detection bounding boxes » Plastic bottle detected [conf: 0.88, bbox: 140,210,48,92] » Paper food wrapper [conf: 0.81, bbox: 320,180,64,40] ✓ Density metric: Moderate litter zone ! Alert state: Logged event for area review
// 01 · The Environmental Question

“Could computer vision help us notice and map environmental waste that humans simply walk past?”

In many busy public spaces — schools, parks, and city transit areas — waste accumulates in blind spots or goes unreported until conditions worsen. Prithvi-ALS explores whether affordable cameras combined with edge computer vision models can detect litter automatically and generate spatial clean-up insights.

// 02 · System Flow

How Prithvi-ALS Works

1

Video Feed

Optical camera monitors designated public or campus zones.

2

Computer Vision

Model detects waste objects (bottles, cups, wrappers).

3

Spatial Logging

Coordinates, timestamps, and waste density are registered.

4

Actionable Insight

Reports and alerts highlight areas needing attention.

// 03 · Development Status

What Has Been Built vs. Explored

Current Stage

Prototyped Functionality

  • Testing camera stream ingestion on personal laptop rig
  • Computer vision model inference with OpenCV & Python
  • Object detection for common litter items (bottles, paper)
  • Basic coordinate bounding-box logging
Under Research

Currently Exploring

  • Lightweight edge model optimization for low-power devices
  • Microcontroller & embedded camera experiments (ESP32-CAM)
  • Temporal filtering to avoid duplicate detection triggers
  • Outdoor lighting and shadow variation robustness
Long-Term Vision

Future Possibilities

  • Fixed campus pilot installations with alert dashboards
  • Integration with smart waste management sorting bins
  • Incentive / reverse-vending recycling machine concepts
  • Environmental clean-up telemetry heatmaps
Loading Prithvi-ALS Vision...