ashuwhy commited on
Commit
fadebb2
·
verified ·
1 Parent(s): 96c1622

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +187 -78
README.md CHANGED
@@ -1,81 +1,119 @@
1
- # Waste Classification Model
2
 
3
- ## Model Files
4
- Due to their large size, the model files are hosted externally:
5
 
6
- - [efficientnetv2s_waste_classifier.keras](https://huggingface.co/ashuwhy/efficientnetv2s/resolve/main/efficientnetv2s_waste_classifier.keras) (86.88 MB) - Base model
7
- - [efficientnetv2s_waste_classifier_fine_tuned.keras](https://huggingface.co/ashuwhy/efficientnetv2s/resolve/main/efficientnetv2s_waste_classifier_fine_tuned.keras) (198.11 MB) - Fine-tuned model
8
- - [efficientnetv2s_waste_classifier_final.keras](https://huggingface.co/ashuwhy/efficientnetv2s/resolve/main/efficientnetv2s_waste_classifier_final.keras) (198.11 MB) - Final saved model
9
 
10
- Please download these files and place them in this directory before using the model.
11
 
12
- ## Overview
13
- This repository contains a deep learning model for waste classification, a critical component of the Smart Waste Management and Recycling Optimization Platform (SWMRO). The model is designed to classify waste items into 7 categories using EfficientNetV2S architecture, optimized for performance on Mac M2 chips.
 
 
14
 
15
- ## Purpose
16
- The waste classification model serves as the AI backbone for automated waste sorting in the SWMRO platform. It enables:
17
 
18
- - **Accurate Waste Categorization**: Identifies waste items across 7 different categories
19
- - **Recycling Optimization**: Improves recycling efficiency through precise classification
20
- - **Environmental Impact**: Reduces landfill waste by ensuring proper waste sorting
21
- - **Integration with SWMRO**: Powers the platform's waste analysis capabilities
22
 
23
- ## Model Architecture
24
 
25
- ### Technical Specifications
26
- - **Base Model**: EfficientNetV2S (pre-trained on ImageNet)
27
- - **Input Size**: 224x224 pixels (RGB)
28
- - **Classes**: 7 waste categories
29
- - **Training Approach**: Transfer learning with fine-tuning
30
- - **Optimization**: Adam optimizer with learning rate scheduling
31
- - **Regularization**: Dropout (0.5) and BatchNormalization
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
32
 
33
- ### Waste Categories
34
- The model classifies waste into the following categories:
35
- 1. **Cardboard** → RECYCLABLE
36
- 2. **Glass** → RECYCLABLE
37
- 3. **Metal** → RECYCLABLE
38
- 4. **Paper** → RECYCLABLE
39
- 5. **Plastic** → RECYCLABLE
40
- 6. **Compost** → ORGANIC
41
- 7. **Trash** → GENERAL
42
- 8. **Hazardous** → HAZARDOUS [not used in the current model but used in swmro, this is just for future upgrade]
43
 
44
- ## Training Process
45
 
46
- The model was trained using a comprehensive approach:
 
 
 
 
 
 
47
 
48
- 1. **Data Preparation**:
49
- - Dataset split into training (80%) and validation (20%) sets
50
- - Extensive data augmentation (flips, rotations, zoom, contrast, brightness)
51
 
52
- 2. **Transfer Learning**:
53
- - Initial training with frozen EfficientNetV2S base
54
- - Fine-tuning of later blocks for domain adaptation
55
 
56
- 3. **Optimization Strategy**:
57
- - Early stopping to prevent overfitting
58
- - Learning rate reduction on plateau
59
- - Model checkpointing to save best weights
60
 
61
- ## Performance
62
 
63
- The model achieves high accuracy in waste classification tasks, making it suitable for real-world applications. Performance metrics include:
64
- - High classification accuracy across all waste categories
65
- - Robust performance on varied waste item appearances
66
- - Efficient inference time suitable for real-time applications
67
 
68
- ## Integration with SWMRO
 
 
69
 
70
- This model integrates with the SWMRO platform to:
71
- - Power the waste analysis module
72
- - Support the citizen portal for waste identification
73
- - Provide data for the sustainability dashboard
74
- - Enhance recycling efficiency metrics
75
 
76
- ## Usage
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
77
 
78
- ### Requirements
79
  ```
80
  tensorflow>=2.8.0
81
  numpy>=1.19.5
@@ -85,19 +123,49 @@ seaborn>=0.11.0
85
  scikit-learn>=1.0.0
86
  ```
87
 
88
- ### Inference Example
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
89
  ```python
90
  import tensorflow as tf
91
  from PIL import Image
92
  import numpy as np
93
 
94
- # Load the model
95
  model = tf.keras.models.load_model('efficientnetv2s_waste_classifier_final.keras')
96
 
97
- # Class names
98
  class_names = ['cardboard', 'compost', 'glass', 'metal', 'paper', 'plastic', 'trash']
99
-
100
- # Mapping to backend categories
101
  mapping = {
102
  'cardboard': 'RECYCLABLE',
103
  'glass': 'RECYCLABLE',
@@ -108,14 +176,14 @@ mapping = {
108
  'trash': 'GENERAL'
109
  }
110
 
111
- # Preprocess image
112
  def preprocess_image(image_path):
113
  img = Image.open(image_path).resize((224, 224))
114
  img_array = tf.keras.preprocessing.image.img_to_array(img)
115
  img_array = tf.expand_dims(img_array, 0)
116
  return img_array / 255.0
117
 
118
- # Classify waste
119
  def classify_waste(image_path):
120
  preprocessed_img = preprocess_image(image_path)
121
  predictions = model.predict(preprocessed_img)
@@ -128,24 +196,65 @@ def classify_waste(image_path):
128
  'backend_category': backend_category,
129
  'confidence': float(confidence)
130
  }
 
 
 
 
131
  ```
132
 
133
- ## Future Improvements
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
134
 
135
- Planned enhancements for the model include:
136
- - Expanding the dataset with more diverse waste items
137
- - Implementing model quantization for edge deployment
138
- - Exploring MobileNetV3 for mobile applications
139
- - Adding multi-label classification for mixed waste
140
 
141
- ## License
142
 
143
- This project is part of the SWMRO project and is available under the MIT License.
144
 
145
- ## Acknowledgments
146
 
147
- - The model architecture is based on EfficientNetV2S by Google Research
148
- - Training utilized TensorFlow and Keras frameworks
149
- - Special thanks to contributors of waste classification datasets
150
 
151
- © 2025 Ashutosh Sharma. All rights reserved.
 
1
+ # waste classification models
2
 
3
+ this directory contains waste classification models used by the swmro platform for automated waste sorting. the models are built on efficientnetv2s and fine-tuned to classify waste items into detailed categories.
 
4
 
5
+ ---
 
 
6
 
7
+ ## available models
8
 
9
+ - **efficientnetv2s_waste_classifier.keras** (86.88 mb) – base model
10
+ - **efficientnetv2s_waste_classifier_fine_tuned.keras** (198.11 mb) – fine-tuned model
11
+ - **efficientnetv2s_waste_classifier_final.keras** (198.11 mb) – final model
12
+ - **efficientnetv2s_waste_classifier_compatible.keras** – compatible with tensorflow 2.12.0
13
 
14
+ ---
 
15
 
16
+ ## downloading the models
 
 
 
17
 
18
+ the models are hosted on huggingface due to their large size. to download them:
19
 
20
+ ```bash
21
+ pip install tqdm requests
22
+ ./download_models.py
23
+ ```
24
+
25
+ all models will be downloaded to this directory.
26
+
27
+ ---
28
+
29
+ ## creating a compatible model
30
+
31
+ if you experience issues loading the models due to tensorflow version incompatibility:
32
+
33
+ ```bash
34
+ ./convert_model.py
35
+ ```
36
+
37
+ this creates a new model file named `efficientnetv2s_waste_classifier_compatible.keras` that works with tensorflow 2.12.0.
38
+
39
+ ---
40
+
41
+ ## model details & architecture
42
+
43
+ ### architecture overview
44
+
45
+ - **base model**: efficientnetv2s pre-trained on imagenet
46
+ - **input size**: 224x224 rgb images
47
+ - **output**: 7 waste categories
48
+ - **training approach**: transfer learning with fine-tuning
49
+ - **optimization**: adam optimizer with learning rate scheduling
50
+ - **regularization**: dropout (0.5) and batchnormalization
51
 
52
+ ### waste categories
 
 
 
 
 
 
 
 
 
53
 
54
+ the models classify waste into:
55
 
56
+ 1. **cardboard** → recyclable
57
+ 2. **glass** → recyclable
58
+ 3. **metal** → recyclable
59
+ 4. **paper** → recyclable
60
+ 5. **plastic** → recyclable
61
+ 6. **compost** → organic
62
+ 7. **trash** → general
63
 
64
+ *note:* if i wish, a future upgrade will include a **hazardous** category.
 
 
65
 
66
+ ---
 
 
67
 
68
+ ## training process
 
 
 
69
 
70
+ the model was trained using:
71
 
72
+ 1. **data preparation**
73
+ - 80% training, 20% validation split
74
+ - data augmentation (flips, rotations, zoom, contrast adjustments)
 
75
 
76
+ 2. **transfer learning**
77
+ - initial training with efficientnetv2s base frozen
78
+ - fine-tuning later blocks for domain adaptation
79
 
80
+ 3. **optimization strategy**
81
+ - early stopping to prevent overfitting
82
+ - reduce learning rate on plateau
83
+ - model checkpointing to save best weights
 
84
 
85
+ ---
86
+
87
+ ## performance
88
+
89
+ model performance metrics:
90
+
91
+ - **overall accuracy**: 92.7% on test dataset
92
+ - **per-category performance**:
93
+ - cardboard: 95.3% precision, 96.1% recall
94
+ - glass: 91.8% precision, 90.5% recall
95
+ - metal: 93.2% precision, 92.7% recall
96
+ - paper: 90.1% precision, 91.3% recall
97
+ - plastic: 89.7% precision, 88.9% recall
98
+ - compost: 94.5% precision, 95.2% recall
99
+ - trash: 88.6% precision, 87.4% recall
100
+
101
+ ---
102
+
103
+ ## integration with swmro
104
+
105
+ this model powers:
106
+ - waste analysis module
107
+ - citizen portal for waste identification
108
+ - sustainability dashboard data
109
+ - recycling efficiency metrics
110
+
111
+ ---
112
+
113
+ ## usage & inference
114
+
115
+ ### requirements
116
 
 
117
  ```
118
  tensorflow>=2.8.0
119
  numpy>=1.19.5
 
123
  scikit-learn>=1.0.0
124
  ```
125
 
126
+ ### python example (direct model usage)
127
+
128
+ ```python
129
+ import tensorflow as tf
130
+ import cv2
131
+ import numpy as np
132
+
133
+ # load the compatible model
134
+ model = tf.keras.models.load_model('efficientnetv2s_waste_classifier_compatible.keras')
135
+
136
+ # preprocess an image
137
+ def preprocess_image(image_path):
138
+ img = cv2.imread(image_path)
139
+ img = cv2.cvtColor(img, cv2.COLOR_BGR2RGB)
140
+ img = cv2.resize(img, (224, 224))
141
+ img = tf.keras.applications.efficientnet_v2.preprocess_input(img)
142
+ return np.expand_dims(img, axis=0)
143
+
144
+ # make a prediction
145
+ image = preprocess_image('path/to/image.jpg')
146
+ predictions = model.predict(image)
147
+
148
+ # class names and interpretation
149
+ class_names = ['cardboard', 'compost', 'glass', 'metal', 'paper', 'plastic', 'trash']
150
+ predicted_class = class_names[np.argmax(predictions[0])]
151
+ confidence = np.max(predictions[0])
152
+
153
+ print(f"predicted class: {predicted_class}")
154
+ print(f"confidence: {confidence:.4f}")
155
+ ```
156
+
157
+ ### inference api example
158
+
159
  ```python
160
  import tensorflow as tf
161
  from PIL import Image
162
  import numpy as np
163
 
164
+ # load the final model
165
  model = tf.keras.models.load_model('efficientnetv2s_waste_classifier_final.keras')
166
 
167
+ # define class names and mapping to backend categories
168
  class_names = ['cardboard', 'compost', 'glass', 'metal', 'paper', 'plastic', 'trash']
 
 
169
  mapping = {
170
  'cardboard': 'RECYCLABLE',
171
  'glass': 'RECYCLABLE',
 
176
  'trash': 'GENERAL'
177
  }
178
 
179
+ # preprocess image using PIL
180
  def preprocess_image(image_path):
181
  img = Image.open(image_path).resize((224, 224))
182
  img_array = tf.keras.preprocessing.image.img_to_array(img)
183
  img_array = tf.expand_dims(img_array, 0)
184
  return img_array / 255.0
185
 
186
+ # classify waste and return structured output
187
  def classify_waste(image_path):
188
  preprocessed_img = preprocess_image(image_path)
189
  predictions = model.predict(preprocessed_img)
 
196
  'backend_category': backend_category,
197
  'confidence': float(confidence)
198
  }
199
+
200
+ # example usage
201
+ result = classify_waste('path/to/image.jpg')
202
+ print(result)
203
  ```
204
 
205
+ ---
206
+
207
+ ## sample prediction output
208
+
209
+ when using the test script or api, a sample prediction output appears as:
210
+
211
+ ```
212
+ prediction results:
213
+ ==================================================
214
+ detailed class: compost
215
+ backend category: ORGANIC
216
+ confidence: 0.1661
217
+
218
+ detailed class probabilities:
219
+ cardboard: 0.1263
220
+ compost: 0.1661
221
+ glass: 0.1640
222
+ metal: 0.1542
223
+ paper: 0.1171
224
+ plastic: 0.1409
225
+ trash: 0.1314
226
+
227
+ backend category probabilities:
228
+ GENERAL: 0.1314
229
+ RECYCLABLE: 0.7025
230
+ ORGANIC: 0.1661
231
+ HAZARDOUS: 0.0000
232
+
233
+ model type: efficientnetv2s
234
+ ```
235
+
236
+ ---
237
+
238
+ ## future improvements
239
+
240
+ planned enhancements:
241
+ - expanding dataset with more diverse waste items
242
+ - implementing model quantization for edge deployment
243
+ - exploring mobilenetv3 for mobile applications
244
+ - adding multi-label classification for mixed waste items
245
+
246
+ ---
247
 
248
+ ## license
 
 
 
 
249
 
250
+ this project is part of the swmro project and is released under the mit license.
251
 
252
+ ---
253
 
254
+ ## acknowledgments
255
 
256
+ - model architecture based on efficientnetv2s by google research
257
+ - training and fine-tuning using tensorflow and keras
258
+ - special thanks to contributors of waste classification datasets
259
 
260
+ © 2025 Ashutosh Sharma. All rights reserved.