iany-waste-v1

An on-device waste-material classifier — point a camera at an item and get its material. A MobileNetV2 trained from open waste datasets, exported to ONNX to run in the browser via onnxruntime-web. Built for iAny, the offline, on-device Khmer AI platform.

Live now: try it at iany.app/waste-scan — fully on-device, nothing uploaded. Help improve it by contributing photos at iany.app/waste.

What it does

Classifies a single item's material into 7 types, for recycling education, correct sorting, and knowing what a waste-buyer will take.

Labels (output order — this order matters)

0 can
1 glass
2 organic
3 other
4 paper
5 plastic_bottle
6 plastic_other

labels.txt in this repo has the same order. Output is a softmax over these 7 classes.

Input / preprocessing (important)

• Input: float32, shape [1, 224, 224, 3] (NHWC).
• Normalization: MobileNetV2 — scale pixels [0,255] → [-1,1] (i.e. x/127.5 - 1).
• Center-crop the frame to a square before resizing to 224×224 for best results.

Usage

Python (onnxruntime)

import onnxruntime as ort, numpy as np
from PIL import Image

labels = ["can","glass","organic","other","paper","plastic_bottle","plastic_other"]
img = Image.open("item.jpg").convert("RGB").resize((224, 224))
x = (np.asarray(img, np.float32) / 127.5 - 1.0)[None]      # [1,224,224,3], [-1,1]
sess = ort.InferenceSession("model.onnx")
probs = sess.run(None, {sess.get_inputs()[0].name: x})[0][0]
print(labels[int(probs.argmax())], float(probs.max()))

Browser (onnxruntime-web) — see src/lib/wasteOnnx.ts in iAny for a live-camera implementation.

Training data

Bootstrapped from open datasets:

• TrashNet (MIT)
• Drinking Waste Classification (Kaggle) — bottle / can / glass / HDPE
• techsash/waste-classification-data (Kaggle) — Organic images only

Base: MobileNetV2 (ImageNet weights). Trained with transfer learning (see the recipe below).

Limitations

• v1 / beta. Trained mostly on Western datasets — accuracy on Cambodian items, brands, and messy real litter is rougher. This improves as /waste photos are folded in and the model is retrained.
• No ewaste class yet (not enough e-waste training images) — 7 of iAny's 8 material types.
• Best on one item filling the frame, decent light. It's a suggestion, not an authoritative sorting decision.

Intended use

Recycling education and sorting guidance; the /waste-scan experiment; and pre-filling labels in the /waste data collector. Not a certified sorting or compliance system.

License & attribution

Released under CC-BY-4.0 — please credit the source datasets (TrashNet — MIT; others per their Kaggle terms) and iAny. Verify each source dataset's terms before commercial redistribution.

Credit & recipe

Trained and released by iAny (E-KHMER Technology). Full training + deploy recipe: github.com/sengtha/iAny · docs/WASTE-MODEL.md.