Hugging Face's logo

jojo-ai-mst
/
burmese_snake_classifier

Keras
TF-Keras
Model card Files Community
2
burmese_snake_classifier / README.md
jojo-ai-mst's picture
jojo-ai-mst
docs: Custom README
9b671fd
preview code
|
raw
history blame
No virus
3.42 kB
metadata
library_name: keras
tags:
  - burmese
  - burma
  - myanmar
  - snake
  - classifier

Model description

MM DeepSnake is an artificial intelligence project to classify snake species in Myanmar. We collect images all around Myanmar for training our model.

Current Version - Alpha - 1.0.0

Currently our model can understand 10 species of snakes. Some of the snakes are very much in species and hard to classify individual species. Therefore, we took genus as a categories.

At the moment, we support

  • Trimeresurus_sp (Asian Palm Pit vipers) - မြွေစိမ်းမြီးခြောက်
  • Rhadophis helleri (Heller Red necked keelback) - လည်ပင်းနီမြွေ
  • Lycodon aulicus (Wolf Snake) - မြွေဝံပုလွေ
  • Fowlea piscator (Checkered Keelback) - ရေမြွေဗျောက်မ
  • Daboia siamensis (Eastern Russell's viper) - မြွေပွေး
  • Chrysopelea ornata (Golden Tree Snake) - ထန်းမြွေ
  • Bungarus fasciatus (Banded Krait) - ငန်းတော်ကြား
  • Ophiophagus hannah(King Cobra) - တောကြီးမြွေဟောက်
  • Laticauda colubrina (Sea Snake) - ဂျက်မြွေ
  • Naja kaouthia (Cobra) - မြွေဟောက်

Here is sample code to use burmese_snake_classifier

import numpy as np
import tensorflow as tf
from huggingface_hub import from_pretrained_keras

pretrained_model = from_pretrained_keras('jojo-ai-mst/burmese_snake_classifier')

class_names = ['Bungarus fasciatus (Banded Krait)', 'Chrysopelea ornata (Golden Tree Snake)', "Daboia siamensis (Eastern Russell's viper)", 'Fowlea piscator (Checkered Keelback)', 'Laticauda colubrina (Sea Snake)', 'Lycodon aulicus (Wolf Snake)', 'Naja kaouthia(Cobra)', 'Ophiophagus_hannah(King Cobra)', 'Rhadophis helleri (Heller Red necked keelback)', 'Trimeresurus_sp (Asian Palm Pit vipers)']

def softmax_stable(x):
    return(np.exp(x - np.max(x)) / np.exp(x - np.max(x)).sum())

def predict_img(input_img):
  img_array = np.expand_dims(input_img, 0) 
  predictions = pretrained_model.predict(img_array)
  score = tf.nn.softmax(predictions[0])

  result =  "This image most likely belongs to {} with a {:.2f} percent confidence.".format(class_names[np.argmax(score)], 100 * np.max(score))

  return result

Intended uses & limitations

This model is open source for open source projects. Project that modifies, extends, derives from this model must mention the original model jojo-ai-mst/burmese_snake_classifier. Commercial use needs to be requested to the model contributor jojo-ai-mst.

We strongly alert that every snake bite case should go to professional medical staffs.

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

Hyperparameters Value
name Adam
weight_decay None
clipnorm None
global_clipnorm None
clipvalue None
use_ema False
ema_momentum 0.99
ema_overwrite_frequency None
jit_compile False
is_legacy_optimizer False
learning_rate 9.999999747378752e-06
beta_1 0.9
beta_2 0.999
epsilon 1e-07
amsgrad False
training_precision float32

Model Plot

View Model Plot

Model Image