grocery_classification

Model Card for Model ID

<!-- Provide a quick summary of what the model is/does. -->

This modelcard aims to be a base template for new models. It has been generated using this raw template.

Model Details

Model Description

<!-- Provide a longer summary of what this model is. -->

• Developed by: Yukieos
• Model type: Image classification with OCR fallback pipeline
• Language(s) (NLP): en
• License: apache-2.0
• Finetuned from model [optional]: torchvision/mobilenet_v3_small

Uses

<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->

Direct Use

<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->

Call the infer_category function to get a grocery label from an image, using OCR first and falling back to a MobileNetV3‐based classifier: category, ocr_text, method = infer_category(image_path) #can replace "image_path" with an address of image query = ocr_text if method=='ocr' else category if method=='manual': query = input("Please type the product you want to search：") print(f"Result: {category} (via {method})")

Out-of-Scope Use

<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->

Non-grocery products (e.g. electronics, clothing)

Packaging text in languages other than English

Highly distorted or tiny product labels

Bias, Risks, and Limitations

<!-- This section is meant to convey both technical and sociotechnical limitations. -->

Class bias: only predicts the ~20 grocery categories seen in training

OCR bias: EasyOCR may misread stylized fonts, low-contrast text, or cluttered backgrounds

Failure mode: when OCR and classifier both low-confidence, returns None

Recommendations

<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->

Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.

How to Get Started with the Model

Use the code below to get started with the model:

pip install torch torchvision easyocr opencv-python pillow

from infer import infer_category

label, raw_text, method = infer_category("image.jpg")

print(f"Label: {label}, OCR saw: {raw_text}, Method: {method}")

Training Details

Training Data

<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->

~2,000 images collected in local grocery stores, evenly split across 20 product categories (e.g. “apple,” “banana,” “milk,” …). Manually labeled and organized into train/val/test splits.

Training Procedure

<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->

Preprocessing

Resize to 224×224 Random horizontal flips, random rotation ±15° Normalize with ImageNet means/stds

Training Hyperparameters

• Training regime: 20 epochs on train split
• Optimizer & LR: Adam, lr = 1e-4
• Batch size: 32 <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->

Evaluation

<!-- This section describes the evaluation protocols and provides the results. -->

Testing Data, Factors & Metrics

Testing Data

<!-- This should link to a Dataset Card if possible. -->

400 held-out images (20 per class) from the same data distribution.

Factors

<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->

Varying lighting Text printed vs. handwritten Partial occlusion

Metrics

<!-- These are the evaluation metrics being used, ideally with a description of why. -->

• Top-1 classification accuracy: 91.2%
• OCR detection precision: 84.5%
• Fallback rate: 12% (cases where OCR matched label, bypassing classifier)

Results

Overall pipeline accuracy on test set: 89.0%

Summary

Environmental Impact

<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->

Carbon emissions can be estimated using the Machine Learning Impact calculator presented in Lacoste et al. (2019).

• Hardware Type: NVIDIA Tesla T4
• Hours used: ~1.5 hours
• Cloud Provider: Google Colab
• Compute Region: us-central1
• Carbon Emitted: ~0.5g

Model Architecture and Objective

Backbone: MobileNetV3-Small, final fc → 20 classes

OCR: EasyOCR reader (en) for text detection + recognition

Loss: CrossEntropy for classification

Compute Infrastructure

Training environment: Colab notebook with single GPU

Inference environment: any machine with PyTorch + EasyOCR

Hardware

NVIDIA T4 (training & eval)

Software

Python 3.10

PyTorch 1.13, torchvision 0.14

EasyOCR 1.4, OpenCV 4.x, Pillow 9.x

Model Card Contact

[email protected]