decimer 2.5.0__tar.gz → 2.7.0__tar.gz

{decimer-2.5.0 → decimer-2.7.0}/DECIMER/DECIMER_EfficinetNetV2_Transfomer_Trainer.py

@@ -76,8 +76,8 @@ AUTO = tf.data.experimental.AUTOTUNE
 
 
 def decode_image(image_data):
-    """Preprocess the input image for Efficient-Net and
-    returned the preprocessed image
+    """Preprocess the input image for Efficient-Net and returned the
+    preprocessed image.
 
     Args:
         image_data (int array): Decoded image in 2D array
@@ -98,8 +98,8 @@ def decode_image(image_data):
 
 
 def read_tfrecord(example):
-    """Read a tf record file and decodes the image and text data
-    back into original form.
+    """Read a tf record file and decodes the image and text data back into
+    original form.
 
     Args:
         example (tf.record): single entry from tf record file
@@ -124,7 +124,7 @@ numbers = re.compile(r"(\d+)")
 
 
 def numericalSort(value):
-    """Sorts the filenames numerically
+    """Sorts the filenames numerically.
 
     Args:
         value (int): numerical value of the file name
@@ -138,7 +138,7 @@ def numericalSort(value):
 
 
 def get_dataset(batch_size=BATCH_SIZE, buffered_size=BUFFER_SIZE, path=""):
-    """Creates a batch of data rom a given dataset
+    """Creates a batch of data rom a given dataset.
 
     Args:
         batch_size (int, optional): number of datapoints per batch. Defaults to BATCH_SIZE.
@@ -207,8 +207,8 @@ print("Training preparation\n")
 
 
 def prepare_for_training(lr_config, encoder_config, transformer_config, verbose=0):
-    """Preparte the model for training. initiate the learning rate, loss object, metrics
-    and optimizer
+    """Preparte the model for training. initiate the learning rate, loss
+    object, metrics and optimizer.
 
     Args:
         lr_config (int): values for learning rate configuration
@@ -239,20 +239,18 @@ def prepare_for_training(lr_config, encoder_config, transformer_config, verbose=
         train_accuracy = tf.keras.metrics.SparseCategoricalAccuracy(
             name="train_accuracy", dtype=tf.float32
         )
-        """
-        validation_loss = tf.keras.metrics.Mean(
-            name="validation_loss", dtype=tf.float32
-        )
-        validation_accuracy = tf.keras.metrics.SparseCategoricalAccuracy(
-            name="validation_accuracy", dtype=tf.float32
-        )
+        """validation_loss = tf.keras.metrics.Mean(
+
+        name="validation_loss", dtype=tf.float32 ) validation_accuracy =
+        tf.keras.metrics.SparseCategoricalAccuracy(
+        name="validation_accuracy", dtype=tf.float32 )
         """
         # Declare the learning rate schedule (try this as actual lr schedule and list...)
         lr_scheduler = config.CustomSchedule(
             transformer_config["d_model"], lr_config["warm_steps"]
         )
 
-        # Instiate an optimizer
+        # Instantiate an optimizer
         optimizer = tf.keras.optimizers.Adam(
             lr_scheduler, beta_1=0.9, beta_2=0.98, epsilon=1e-9
         )
@@ -325,9 +323,9 @@ if ckpt_manager.latest_checkpoint:
     start_epoch = int(ckpt_manager.latest_checkpoint.split("-")[-1])
 
 
-# Main training step fucntion
+# Main training step function
 def train_step(image_batch, selfies_batch):
-    """Main training step function
+    """Main training step function.
 
     Args:
         image_batch (float array): Input image batch
@@ -378,7 +376,7 @@ def dist_train_step(image_batch, selfies_batch):
 
 
 """
-# Main validation step fucntion
+# Main validation step function
 def validation_step(image_batch, selfies_batch):
 
     selfies_batch_input = selfies_batch[:, :-1]

{decimer-2.5.0 → decimer-2.7.0}/DECIMER/Predictor_usingCheckpoints.py

@@ -45,7 +45,7 @@ tokenizer = pickle.load(
 
 max_length = 302
 
-# Image partameters
+# Image parameters
 IMG_EMB_DIM = (16, 16, 512)
 IMG_EMB_DIM = (IMG_EMB_DIM[0] * IMG_EMB_DIM[1], IMG_EMB_DIM[2])
 IMG_SHAPE = (512, 512, 3)
@@ -130,9 +130,8 @@ class DECIMER_Predictor(tf.Module):
         self.max_length = max_length
 
     def __call__(self, Decoded_image):
-        """
-        Run the DECIMER predictor model when called.
-        Usage of predict_SMILES or predict_SMILES_with_confidence is recommended instead
+        """Run the DECIMER predictor model when called. Usage of predict_SMILES
+        or predict_SMILES_with_confidence is recommended instead.
 
         Args:
             Decoded_image (_type_): output of config.decode_image
@@ -180,9 +179,8 @@ class DECIMER_Predictor(tf.Module):
 
 
 def detokenize_output(predicted_array: tf.Tensor) -> str:
-    """
-    This function takes the predicted array of tokens and returns the predicted SMILES
-    string.
+    """This function takes the predicted array of tokens and returns the
+    predicted SMILES string.
 
     Args:
         predicted_array (tf.Tensor): Transformer Decoder output array (predicted tokens)
@@ -203,11 +201,10 @@ def detokenize_output_add_confidence(
     predicted_array: tf.Tensor,
     confidence_array: tf.Tensor,
 ) -> List[Tuple[str, float]]:
-    """
-    This function takes the predicted array of tokens as well as the confidence values
-    returned by the Transformer Decoder and returns a list of tuples
-    that contain each token of the predicted SMILES string and the confidence
-    value.
+    """This function takes the predicted array of tokens as well as the
+    confidence values returned by the Transformer Decoder and returns a list of
+    tuples that contain each token of the predicted SMILES string and the
+    confidence value.
 
     Args:
         predicted_array (tf.Tensor): Transformer Decoder output array (predicted tokens)
@@ -234,8 +231,7 @@ DECIMER = DECIMER_Predictor(encoder, tokenizer, transformer, MAX_LEN)
 
 
 def predict_SMILES_with_confidence(image_path: str) -> List[Tuple[str, float]]:
-    """
-    This function takes an image path (str) and returns a list of tuples
+    """This function takes an image path (str) and returns a list of tuples
     that contain each token of the predicted SMILES string and the confidence
     level from the last layer of the Transformer decoder.
 
@@ -255,8 +251,7 @@ def predict_SMILES_with_confidence(image_path: str) -> List[Tuple[str, float]]:
 
 
 def predict_SMILES(image_path: str) -> str:
-    """
-    This function takes an image path (str) and returns the SMILES
+    """This function takes an image path (str) and returns the SMILES
     representation of the depicted molecule (str).
 
     Args:

{decimer-2.5.0 → decimer-2.7.0}/DECIMER/Repack_model.py

@@ -87,9 +87,8 @@ if ckpt_manager.latest_checkpoint:
 
 
 def detokenize_output(predicted_array: tf.Tensor) -> str:
-    """
-    This function takes the predicted array of tokens and returns the predicted SMILES
-    string.
+    """This function takes the predicted array of tokens and returns the
+    predicted SMILES string.
 
     Args:
         predicted_array (tf.Tensor): Transformer Decoder output array (predicted tokens)
@@ -110,11 +109,10 @@ def detokenize_output_add_confidence(
     predicted_array: tf.Tensor,
     confidence_array: tf.Tensor,
 ) -> List[Tuple[str, float]]:
-    """
-    This function takes the predicted array of tokens as well as the confidence values
-    returned by the Transformer Decoder and returns a list of tuples
-    that contain each token of the predicted SMILES string and the confidence
-    value.
+    """This function takes the predicted array of tokens as well as the
+    confidence values returned by the Transformer Decoder and returns a list of
+    tuples that contain each token of the predicted SMILES string and the
+    confidence value.
 
     Args:
         predicted_array (tf.Tensor): Transformer Decoder output array (predicted tokens)
@@ -137,18 +135,22 @@ def detokenize_output_add_confidence(
 
 
 class DECIMER_Predictor(tf.Module):
-    """This is a class which takes care of inference. It loads the saved checkpoint and the necessary
-    tokenizers. The inference begins with the start token (<start>) and ends when the end token(<end>)
-    is met. This class can only work with tf.Tensor objects. The strings shoul gets transformed into np.arrays
-    before feeding them into this class.
+    """This is a class which takes care of inference.
+
+    It loads the saved checkpoint and the necessary tokenizers. The
+    inference begins with the start token (<start>) and ends when the
+    end token(<end>) is met. This class can only work with tf.Tensor
+    objects. The strings should get transformed into np.arrays before
+    feeding them into this class.
     """
 
     def __init__(self, encoder, tokenizer, transformer, max_length):
-        """Load the tokenizers, the maximum input and output length and the model.
+        """Load the tokenizers, the maximum input and output length and the
+        model.
 
         Args:
             encoder (tf.keras.model):  The encoder model
-            tokenizer (tf.keras.tokenizer): Output tokenizer, defines which charater is assigned to what token
+            tokenizer (tf.keras.tokenizer): Output tokenizer, defines which character is assigned to what token
             transformer (tf.keras.model):  The transformer model
             max_length (int): Maximum length of a string which can get predicted
         """
@@ -158,8 +160,8 @@ class DECIMER_Predictor(tf.Module):
         self.max_length = max_length
 
     def __call__(self, Decoded_image):
-        """This fuction takes in the Decoded image as input and
-        makes the predicted list of tokens and return the tokens as tf.Tensor array.
+        """This function takes in the Decoded image as input and makes the
+        predicted list of tokens and return the tokens as tf.Tensor array.
         Before feeding the input array we must define start and the end tokens.
 
         Args:
@@ -211,7 +213,9 @@ DECIMER = DECIMER_Predictor(encoder, tokenizer, transformer, MAX_LEN)
 
 
 class ExportDECIMERPredictor(tf.Module):
-    """This class wraps the inference class into a module into tf.Module sub-class, with a tf.function on the __call__ method.
+    """This class wraps the inference class into a module into tf.Module sub-
+    class, with a tf.function on the __call__ method.
+
     So we could export the model as a tf.saved_model.
     """
 
@@ -221,9 +225,10 @@ class ExportDECIMERPredictor(tf.Module):
 
     @tf.function
     def __call__(self, Decoded_Image):
-        """This fucntion calls the __call__function from the translator class.
-        In the tf.function only the output sentence is returned.
-        Thanks to the non-strict execution in tf.function any unnecessary values are never computed.
+        """This function calls the __call__function from the translator class.
+        In the tf.function only the output sentence is returned. Thanks to the
+        non-strict execution in tf.function any unnecessary values are never
+        computed.
 
         Args:
             sentence (tf.Tensor[tf.int32]): Input array in tf.Easgertensor format.

{decimer-2.5.0 → decimer-2.7.0}/DECIMER/Utils/Create_image_tokenizer.py

@@ -70,7 +70,7 @@ def data_loader(Smiles_Path):
     all_img_name = []
 
     for line in smiles.split("\n"):
-        # Split the ID and SMILES to seperate tokens
+        # Split the ID and SMILES to separate tokens
         tokens = line.split(",")
 
         image_id = str(tokens[0]) + ".png"

{decimer-2.5.0 → decimer-2.7.0}/DECIMER/__init__.py

@@ -1,32 +1,29 @@
-# -*- coding: utf-8 -*-
-"""
-DECIMER V2.4.0 Python Package.
-============================
-
-This repository contains DECIMER-V2,Deep lEarning for Chemical ImagE Recognition) project
-was launched to address the OCSR problem with the latest computational intelligence methods
-to provide an automated open-source software solution.
-
-
-Typical usage example:
-
-from decimer import predict_SMILES
-
-# Chemical depiction to SMILES translation
-image_path = "path/to/imagefile"
-SMILES = predict_SMILES(image_path)
-print(SMILES)
-
-For comments, bug reports or feature ideas,
-please raise a issue on the Github repository.
-
-"""
-
-__version__ = "2.4.0"
-
-__all__ = [
-    "DECIMER",
-]
-
-
-from .decimer import predict_SMILES, predict_SMILES_with_confidence
+# -*- coding: utf-8 -*-
+"""DECIMER V2.6.0 Python Package. ============================
+
+This repository contains DECIMER-V2,Deep lEarning for Chemical ImagE Recognition) project
+was launched to address the OCSR problem with the latest computational intelligence methods
+to provide an automated open-source software solution.
+
+
+Typical usage example:
+
+from decimer import predict_SMILES
+
+# Chemical depiction to SMILES translation
+image_path = "path/to/imagefile"
+SMILES = predict_SMILES(image_path)
+print(SMILES)
+
+For comments, bug reports or feature ideas,
+please raise a issue on the Github repository.
+"""
+
+__version__ = "2.6.0"
+
+__all__ = [
+    "DECIMER",
+]
+
+
+from .decimer import predict_SMILES

{decimer-2.5.0 → decimer-2.7.0}/DECIMER/config.py

@@ -20,9 +20,10 @@ TARGET_DTYPE = tf.float32
 
 
 def resize_byratio(image):
-    """
-    This function takes a Pillow Image object and will resize the image by 512 x 512
-    To upscale or to downscale the image LANCZOS resampling method is used.
+    """This function takes a Pillow Image object and will resize the image by
+    512 x 512 To upscale or to downscale the image LANCZOS resampling method is
+    used.
+
     with the new pillow version the antialias is turned on when using LANCZOS.
     Args: PIL.Image
     Returns: PIL.Image
@@ -35,10 +36,10 @@ def resize_byratio(image):
 
 
 def central_square_image(image):
-    """
-    This function takes a Pillow Image object and will add white padding
-    so that the image has a square shape with the width/height of the longest side
+    """This function takes a Pillow Image object and will add white padding so
+    that the image has a square shape with the width/height of the longest side
     of the original image.
+
     Args: PIL.Image
     Returns: PIL.Image
     """
@@ -59,6 +60,7 @@ def central_square_image(image):
 def delete_empty_borders(image):
     """This function takes a Pillow Image object, converts it to grayscale and
     deletes white space at the borders.
+
     Args: PIL.Image
     Returns: PIL.Image
     """
@@ -132,8 +134,8 @@ def get_bnw_image(image):
 
 
 def increase_contrast(image):
-    """
-    This function increases the contrast of an image input.
+    """This function increases the contrast of an image input.
+
     Args: PIL.Image
     Returns: PIL.Image
     """
@@ -153,8 +155,9 @@ def increase_contrast(image):
 
 
 def get_resize(image):
-    """
-    This function used to decide how to resize a given image without losing much information.
+    """This function used to decide how to resize a given image without losing
+    much information.
+
     Args: PIL.Image
     Returns: PIL.Image
     """
@@ -172,8 +175,8 @@ def get_resize(image):
 
 
 def increase_brightness(image):
-    """
-    This function adjusts the brightness of the given image.
+    """This function adjusts the brightness of the given image.
+
     Args: PIL.Image
     Returns: PIL.Image
     """
@@ -183,8 +186,8 @@ def increase_brightness(image):
 
 
 def decode_image(image_path: str):
-    """
-    Loads an image and preprocesses the input image in several steps to get the image ready for DECIMER input.
+    """Loads an image and preprocesses the input image in several steps to get
+    the image ready for DECIMER input.
 
     Args:
         image_path (str): path of input image
@@ -207,9 +210,7 @@ def decode_image(image_path: str):
 
 
 class Config:
-    """
-    Configuration class
-    """
+    """Configuration class."""
 
     def __init__(
         self,
@@ -227,11 +228,11 @@ class Config:
         do_permute=False,
         pretrained_weights=None,
     ):
-        """This functions initializes the Efficient-Net V2 encoder with user defined
-        configurations.
+        """This functions initializes the Efficient-Net V2 encoder with user
+        defined configurations.
 
         Args:
-            image_embedding_dim (int): Embedding dimention of the input image
+            image_embedding_dim (int): Embedding dimension of the input image
             preprocessing_fn (method): Efficient Net preprocessing function for input image
             backbone_fn (method): Calls Efficient-Net V2 as backbone for encoder
             image_shape (int): Shape of the input image
@@ -257,9 +258,8 @@ class Config:
         image_embedding_dim,
         rate=0.1,
     ):
-        """This functions initializes the Transformer model as decoder with user defined
-        configurations.
-
+        """This functions initializes the Transformer model as decoder with
+        user defined configurations.
 
         Args:
             vocab_len (int): Total number of words in the input vocabulary
@@ -267,7 +267,7 @@ class Config:
             n_transformer_layers (int): Number of layers present in the transformer model
             transformer_d_dff (int): Transformer feed forward upwards projection size
             transformer_n_heads (int): Number of heads present in the transformer model
-            image_embedding_dim (int): Total number of dimension the image gets embeddeded
+            image_embedding_dim (int): Total number of dimension the image gets embedded
             dropout_rate (float, optional): Fraction of the input units to drop. Defaults to 0.1.
         """
         self.transformer_config = dict(
@@ -281,7 +281,7 @@ class Config:
         )
 
     def initialize_lr_config(self, warm_steps, n_epochs):
-        """This function sets the configuration to initialize learning rate
+        """This function sets the configuration to initialize learning rate.
 
         Args:
             warm_steps (int): Number of steps The learning rate is increased
@@ -294,7 +294,7 @@ class Config:
 
 
 class CustomSchedule(tf.keras.optimizers.schedules.LearningRateSchedule):
-    """Custom schedule for learning rate used during training
+    """Custom schedule for learning rate used during training.
 
     Args:
         tf (_type_): keras learning rate schedule
@@ -316,9 +316,9 @@ class CustomSchedule(tf.keras.optimizers.schedules.LearningRateSchedule):
 
 
 def prepare_models(encoder_config, transformer_config, replica_batch_size, verbose=0):
-    """This function is used to initiate the Encoder and the Transformer with appropriate
-    configs set by the user. After initiating the models this function returns the Encoder,Transformer
-    and the optimizer.
+    """This function is used to initiate the Encoder and the Transformer with
+    appropriate configs set by the user. After initiating the models this
+    function returns the Encoder,Transformer and the optimizer.
 
     Args:
         encoder_config ([type]): Encoder configuration set by user in the config class.
@@ -330,7 +330,7 @@ def prepare_models(encoder_config, transformer_config, replica_batch_size, verbo
         [type]: Optimizer, Encoder model and the Transformer
     """
 
-    # Instiate an optimizer
+    # Instantiate an optimizer
     optimizer = tf.keras.optimizers.Adam(learning_rate=0.00051)
 
     # Instantiate the encoder model
@@ -353,9 +353,10 @@ def prepare_models(encoder_config, transformer_config, replica_batch_size, verbo
 
 # Downloads the model and unzips the file downloaded, if the model is not present on the working directory.
 def download_trained_weights(model_url: str, model_path: str, verbose=1):
-    """This function downloads the trained models and tokenizers to a default location.
-    After downloading the zipped file the function unzips the file automatically.
-    If the model exists on the default location this function will not work.
+    """This function downloads the trained models and tokenizers to a default
+    location. After downloading the zipped file the function unzips the file
+    automatically. If the model exists on the default location this function
+    will not work.
 
     Args:
         model_url (str): trained model url for downloading.

decimer-2.7.0/DECIMER/decimer.py

@@ -0,0 +1,171 @@
+import logging
+import os
+import pickle
+import sys
+from typing import List
+from typing import Tuple
+
+import pystow
+import tensorflow as tf
+
+import DECIMER.config as config
+import DECIMER.utils as utils
+
+# Silence tensorflow model loading warnings.
+logging.getLogger("absl").setLevel("ERROR")
+
+# Silence tensorflow errors - not recommended if your model is not working properly.
+os.environ["TF_CPP_MIN_LOG_LEVEL"] = "3"
+
+# Set the absolute path
+HERE = os.path.dirname(os.path.abspath(__file__))
+
+# Set model to run on default GPU and allow memory to grow as much as needed.
+# This allows us to run multiple instances of inference in the same GPU.
+os.environ["CUDA_VISIBLE_DEVICES"] = "0"
+gpus = tf.config.experimental.list_physical_devices("GPU")
+for gpu in gpus:
+    tf.config.experimental.set_memory_growth(gpu, True)
+
+# Set path
+default_path = pystow.join("DECIMER-V2")
+
+model_urls = {
+    "DECIMER": "https://zenodo.org/record/8300489/files/models.zip",
+    "DECIMER_HandDrawn": "https://zenodo.org/records/10781330/files/DECIMER_HandDrawn_model.zip",
+}
+
+
+def get_models(model_urls: dict):
+    """Download and load models from the provided URLs.
+
+    This function downloads models from the provided URLs to a default location,
+    then loads tokenizers and TensorFlow saved models.
+
+    Args:
+        model_urls (dict): A dictionary containing model names as keys and their corresponding URLs as values.
+
+    Returns:
+        tuple: A tuple containing loaded tokenizer and TensorFlow saved models.
+            - tokenizer (object): Tokenizer for DECIMER model.
+            - DECIMER_V2 (tf.saved_model): TensorFlow saved model for DECIMER.
+            - DECIMER_Hand_drawn (tf.saved_model): TensorFlow saved model for DECIMER HandDrawn.
+    """
+    # Download models to a default location
+    model_paths = utils.ensure_models(default_path=default_path, model_urls=model_urls)
+
+    # Load tokenizers
+    tokenizer_path = os.path.join(
+        model_paths["DECIMER"], "assets", "tokenizer_SMILES.pkl"
+    )
+    tokenizer = pickle.load(open(tokenizer_path, "rb"))
+
+    # Load DECIMER models
+    DECIMER_V2 = tf.saved_model.load(model_paths["DECIMER"])
+    DECIMER_Hand_drawn = tf.saved_model.load(model_paths["DECIMER_HandDrawn"])
+
+    return tokenizer, DECIMER_V2, DECIMER_Hand_drawn
+
+
+tokenizer, DECIMER_V2, DECIMER_Hand_drawn = get_models(model_urls)
+
+
+def detokenize_output(predicted_array: int) -> str:
+    """This function takes the predicted tokens from the DECIMER model and
+    returns the decoded SMILES string.
+
+    Args:
+        predicted_array (int): Predicted tokens from DECIMER
+
+    Returns:
+        (str): SMILES representation of the molecule
+    """
+    outputs = [tokenizer.index_word[i] for i in predicted_array[0].numpy()]
+    prediction = (
+        "".join([str(elem) for elem in outputs])
+        .replace("<start>", "")
+        .replace("<end>", "")
+    )
+    return prediction
+
+
+def detokenize_output_add_confidence(
+    predicted_array: tf.Tensor,
+    confidence_array: tf.Tensor,
+) -> List[Tuple[str, float]]:
+    """This function takes the predicted array of tokens as well as the
+    confidence values returned by the Transformer Decoder and returns a list of
+    tuples that contain each token of the predicted SMILES string and the
+    confidence value.
+
+    Args:
+        predicted_array (tf.Tensor): Transformer Decoder output array (predicted tokens)
+
+    Returns:
+        str: SMILES string
+    """
+    prediction_with_confidence = [
+        (
+            tokenizer.index_word[predicted_array[0].numpy()[i]],
+            confidence_array[i].numpy(),
+        )
+        for i in range(len(confidence_array))
+    ]
+    # remove start and end tokens
+    prediction_with_confidence_ = prediction_with_confidence[1:-1]
+
+    decoded_prediction_with_confidence = list(
+        [(utils.decoder(tok), conf) for tok, conf in prediction_with_confidence_]
+    )
+
+    return decoded_prediction_with_confidence
+
+
+def predict_SMILES(
+    image_path: str, confidence: bool = False, hand_drawn: bool = False
+) -> str:
+    """Predicts SMILES representation of a molecule depicted in the given image.
+
+    Args:
+        image_path (str): Path of chemical structure depiction image
+        confidence (bool): Flag to indicate whether to return confidence values along with SMILES prediction
+        hand_drawn (bool): Flag to indicate whether the molecule in the image is hand-drawn
+
+    Returns:
+        str: SMILES representation of the molecule in the input image, optionally with confidence values
+    """
+    chemical_structure = config.decode_image(image_path)
+
+    model = DECIMER_Hand_drawn if hand_drawn else DECIMER_V2
+    predicted_tokens, confidence_values = model(tf.constant(chemical_structure))
+
+    predicted_SMILES = utils.decoder(detokenize_output(predicted_tokens))
+
+    if confidence:
+        predicted_SMILES_with_confidence = detokenize_output_add_confidence(
+            predicted_tokens, confidence_values
+        )
+        return predicted_SMILES, predicted_SMILES_with_confidence
+
+    return predicted_SMILES
+
+
+def main():
+    """This function take the path of the image as user input and returns the
+    predicted SMILES as output in CLI.
+
+    Agrs:
+        str: image_path
+
+    Returns:
+        str: predicted SMILES
+    """
+    if len(sys.argv) != 2:
+        print("Usage: {} $image_path".format(sys.argv[0]))
+    else:
+        SMILES = predict_SMILES(sys.argv[1])
+        print(SMILES)
+
+
+if __name__ == "__main__":
+    main()

{decimer-2.5.0 → decimer-2.7.0}/DECIMER/efficientnetv2/effnetv2_configs.py

@@ -63,7 +63,8 @@ class BlockDecoder(object):
         return "_".join(args)
 
     def decode(self, string_list):
-        """Decodes a list of string notations to specify blocks inside the network.
+        """Decodes a list of string notations to specify blocks inside the
+        network.
 
         Args:
           string_list: a list of strings, each string is a notation of block.

{decimer-2.5.0 → decimer-2.7.0}/DECIMER/efficientnetv2/effnetv2_model.py

@@ -180,7 +180,7 @@ class MBConvBlock(tf.keras.layers.Layer):
 
         self.endpoints = None
 
-        # Builds the block accordings to arguments.
+        # Builds the block according to arguments.
         self._build()
 
     @property

{decimer-2.5.0 → decimer-2.7.0}/DECIMER/efficientnetv2/hparams.py

@@ -158,7 +158,10 @@ class Config(dict):
             yaml.dump(self.as_dict(), f, default_flow_style=False)
 
     def parse_from_str(self, config_str: Text) -> Dict[Any, Any]:
-        """Parse a string like 'x.y=1,x.z=2' to nested dict {x: {y: 1, z: 2}}."""
+        """Parse a string like 'x.y=1,x.z=2' to nested dict {x: {y: 1, z:
+
+        2}}.
+        """
         if not config_str:
             return {}
         config_dict = {}

{decimer-2.5.0 → decimer-2.7.0}/DECIMER/efficientnetv2/utils.py

@@ -192,7 +192,8 @@ class TpuBatchNormalization(tf.keras.layers.BatchNormalization):
         super().__init__(fused=fused, **kwargs)
 
     def _moments(self, inputs, reduction_axes, keep_dims):
-        """Compute the mean and variance: it overrides the original _moments."""
+        """Compute the mean and variance: it overrides the original
+        _moments."""
         shard_mean, shard_variance = super()._moments(
             inputs, reduction_axes, keep_dims=keep_dims
         )
@@ -222,7 +223,8 @@ class TpuBatchNormalization(tf.keras.layers.BatchNormalization):
 
 
 class BatchNormalization(tf.keras.layers.BatchNormalization):
-    """Fixed default name of BatchNormalization to match TpuBatchNormalization."""
+    """Fixed default name of BatchNormalization to match
+    TpuBatchNormalization."""
 
     def __init__(self, **kwargs):
         if not kwargs.get("name", None):
@@ -371,7 +373,10 @@ def get_tpu_host_call(global_step, model_dir, iterations_per_loop):
         return None  # No summaries to write.
 
     def host_call_fn(global_step, *args):
-        """Training host call. Creates summaries for training metrics."""
+        """Training host call.
+
+        Creates summaries for training metrics.
+        """
         gs = global_step[0]
         with tf.summary.create_file_writer(
             model_dir, max_queue=iterations_per_loop
@@ -594,7 +599,8 @@ def restore_tf2_ckpt(model, ckpt_path_or_file, skip_mismatch=True, exclude_layer
 
 
 class ReuableBackupAndRestore(tf.keras.callbacks.experimental.BackupAndRestore):
-    """A BackupAndRestore callback that can be used across multiple model.fit()s."""
+    """A BackupAndRestore callback that can be used across multiple
+    model.fit()s."""
 
     def on_train_end(self, logs=None):
         # don't delete the backup, so it can be used for future model.fit()s

{decimer-2.5.0 → decimer-2.7.0}/DECIMER/utils.py

@@ -63,27 +63,35 @@ def decoder(predictions):
     return modified
 
 
-def ensure_model(
-    default_path: str,
-    model_url: str = "https://zenodo.org/record/8300489/files/models.zip",
-):
-    """Function to ensure model is present locally
+def ensure_models(default_path: str, model_urls: dict) -> dict:
+    """Function to ensure models are present locally.
 
-    Convenient function to ensure model download before usage
+    Convenient function to ensure model downloads before usage
 
     Args:
-        default path (str): default path for DECIMER data
-        model_url (str): trained model url for downloading
-    """
+        default_path (str): Default path for model data
+        model_urls (dict): Dictionary containing model names as keys and their corresponding URLs as values
 
-    model_path = os.path.join(default_path.as_posix(), "DECIMER_model")
-    print(model_path)
-
-    if (
-        os.path.exists(model_path)
-        and os.stat(os.path.join(model_path, "saved_model.pb")).st_size != 28080309
-    ):
-        shutil.rmtree(model_path)
-        config.download_trained_weights(model_url, default_path)
-    elif not os.path.exists(model_path):
-        config.download_trained_weights(model_url, default_path)
+    Returns:
+        dict: A dictionary containing model names as keys and their local paths as values
+    """
+    model_paths = {}
+    # Store st_size of each model
+    model_sizes = {
+        "DECIMER": 28080309,
+        "DECIMER_HandDrawn": 28080328,
+    }
+    for model_name, model_url in model_urls.items():
+        model_path = os.path.join(default_path, f"{model_name}_model")
+        if os.path.exists(model_path) and os.stat(
+            os.path.join(model_path, "saved_model.pb")
+        ).st_size != model_sizes.get(model_name):
+            print(f"Working with model {model_name}")
+            shutil.rmtree(model_path)
+            config.download_trained_weights(model_url, default_path)
+        elif not os.path.exists(model_path):
+            config.download_trained_weights(model_url, default_path)
+
+        # Store the model path
+        model_paths[model_name] = model_path
+    return model_paths

{decimer-2.5.0/decimer.egg-info → decimer-2.7.0}/PKG-INFO

@@ -1,7 +1,7 @@
 Metadata-Version: 2.1
 Name: decimer
-Version: 2.5.0
-Summary: DECIMER 2.4.0: Deep Learning for Chemical Image Recognition using Efficient-Net V2 + Transformer
+Version: 2.7.0
+Summary: DECIMER 2.6.0: Deep Learning for Chemical Image Recognition using Efficient-Net V2 + Transformer
 Home-page: https://github.com/Kohulan/DECIMER-Image_Transformer
 Author: Kohulan Rajan
 Author-email: kohulan.rajan@uni-jena.de
@@ -111,7 +111,7 @@ print(SMILES)
 - This project is licensed under the MIT License - see the [LICENSE](https://raw.githubusercontent.com/Kohulan/DECIMER-Image_Transformer/master/LICENSE?token=AHKLIF3EULMCUKCFUHIPBMDARSMDO) file for details
 
 ## Citation
-- Rajan K, Brinkhaus HO, Agea MI, Zielesny A, Steinbeck C (2023) DECIMER.ai - An open platform for automated optical chemical structure identification, segmentation and recognition in scientific publications. ChemRxiv. doi: https://10.26434/chemrxiv-2023-xhcx9
+- Rajan K, Brinkhaus HO, Agea MI, Zielesny A, Steinbeck C DECIMER.ai - An open platform for automated optical chemical structure identification, segmentation and recognition in scientific publications. Nat. Commun. 14, 5045 (2023). https://doi.org/10.1038/s41467-023-40782-0
 - Rajan, K., Zielesny, A. & Steinbeck, C. DECIMER 1.0: deep learning for chemical image recognition using transformers. J Cheminform 13, 61 (2021). https://doi.org/10.1186/s13321-021-00538-8
 
 ## References

{decimer-2.5.0 → decimer-2.7.0}/README.md

@@ -87,7 +87,7 @@ print(SMILES)
 - This project is licensed under the MIT License - see the [LICENSE](https://raw.githubusercontent.com/Kohulan/DECIMER-Image_Transformer/master/LICENSE?token=AHKLIF3EULMCUKCFUHIPBMDARSMDO) file for details
 
 ## Citation
-- Rajan K, Brinkhaus HO, Agea MI, Zielesny A, Steinbeck C (2023) DECIMER.ai - An open platform for automated optical chemical structure identification, segmentation and recognition in scientific publications. ChemRxiv. doi: https://10.26434/chemrxiv-2023-xhcx9
+- Rajan K, Brinkhaus HO, Agea MI, Zielesny A, Steinbeck C DECIMER.ai - An open platform for automated optical chemical structure identification, segmentation and recognition in scientific publications. Nat. Commun. 14, 5045 (2023). https://doi.org/10.1038/s41467-023-40782-0
 - Rajan, K., Zielesny, A. & Steinbeck, C. DECIMER 1.0: deep learning for chemical image recognition using transformers. J Cheminform 13, 61 (2021). https://doi.org/10.1186/s13321-021-00538-8
 
 ## References

{decimer-2.5.0 → decimer-2.7.0/decimer.egg-info}/PKG-INFO

@@ -1,7 +1,7 @@
 Metadata-Version: 2.1
 Name: decimer
-Version: 2.5.0
-Summary: DECIMER 2.4.0: Deep Learning for Chemical Image Recognition using Efficient-Net V2 + Transformer
+Version: 2.7.0
+Summary: DECIMER 2.6.0: Deep Learning for Chemical Image Recognition using Efficient-Net V2 + Transformer
 Home-page: https://github.com/Kohulan/DECIMER-Image_Transformer
 Author: Kohulan Rajan
 Author-email: kohulan.rajan@uni-jena.de
@@ -111,7 +111,7 @@ print(SMILES)
 - This project is licensed under the MIT License - see the [LICENSE](https://raw.githubusercontent.com/Kohulan/DECIMER-Image_Transformer/master/LICENSE?token=AHKLIF3EULMCUKCFUHIPBMDARSMDO) file for details
 
 ## Citation
-- Rajan K, Brinkhaus HO, Agea MI, Zielesny A, Steinbeck C (2023) DECIMER.ai - An open platform for automated optical chemical structure identification, segmentation and recognition in scientific publications. ChemRxiv. doi: https://10.26434/chemrxiv-2023-xhcx9
+- Rajan K, Brinkhaus HO, Agea MI, Zielesny A, Steinbeck C DECIMER.ai - An open platform for automated optical chemical structure identification, segmentation and recognition in scientific publications. Nat. Commun. 14, 5045 (2023). https://doi.org/10.1038/s41467-023-40782-0
 - Rajan, K., Zielesny, A. & Steinbeck, C. DECIMER 1.0: deep learning for chemical image recognition using transformers. J Cheminform 13, 61 (2021). https://doi.org/10.1186/s13321-021-00538-8
 
 ## References

{decimer-2.5.0 → decimer-2.7.0}/decimer.egg-info/SOURCES.txt

@@ -19,5 +19,6 @@ DECIMER/efficientnetv2/utils.py
 decimer.egg-info/PKG-INFO
 decimer.egg-info/SOURCES.txt
 decimer.egg-info/dependency_links.txt
+decimer.egg-info/entry_points.txt
 decimer.egg-info/requires.txt
 decimer.egg-info/top_level.txt

decimer-2.7.0/decimer.egg-info/entry_points.txt

@@ -0,0 +1,2 @@
+[console_scripts]
+decimer = DECIMER.decimer:main

{decimer-2.5.0 → decimer-2.7.0}/decimer.egg-info/requires.txt

@@ -1,4 +1,4 @@
-tensorflow==2.12.0
+tensorflow<=2.15.0,>=2.12.0
 opencv-python
 pystow
 pillow-heif

{decimer-2.5.0 → decimer-2.7.0}/setup.py

@@ -1,52 +1,55 @@
-#!/usr/bin/env python
-import platform
-
-import setuptools
-
-if (
-    platform.processor() == "arm" or platform.processor() == "i386"
-) and platform.system() == "Darwin":
-    tensorflow_os = "tensorflow-macos>=2.10.0"
-else:
-    tensorflow_os = "tensorflow==2.12.0"
-
-with open("README.md", "r") as fh:
-    long_description = fh.read()
-
-setuptools.setup(
-    name="decimer",
-    version="2.5.0",
-    author="Kohulan Rajan",
-    author_email="kohulan.rajan@uni-jena.de",
-    maintainer="Kohulan Rajan, Otto Brinkhaus ",
-    maintainer_email="kohulan.rajan@uni-jena.de, otto.brinkhaus@uni-jena.de",
-    description="DECIMER 2.4.0: Deep Learning for Chemical Image Recognition using Efficient-Net V2 + Transformer",
-    long_description=long_description,
-    long_description_content_type="text/markdown",
-    url="https://github.com/Kohulan/DECIMER-Image_Transformer",
-    packages=setuptools.find_packages(),
-    license="MIT",
-    install_requires=[
-        tensorflow_os,
-        "opencv-python",
-        "pystow",
-        "pillow-heif",
-        "efficientnet",
-        "selfies",
-        "pyyaml",
-    ],
-    package_data={"DECIMER": ["repack/*.*", "efficientnetv2/*.*", "Utils/*.*"]},
-    classifiers=[
-        "Programming Language :: Python :: 3",
-        "Programming Language :: Python :: 3.5",
-        "Programming Language :: Python :: 3.6",
-        "Programming Language :: Python :: 3.7",
-        "Programming Language :: Python :: 3.8",
-        "Programming Language :: Python :: 3.9",
-        "Programming Language :: Python :: 3.10",
-        "Programming Language :: Python :: 3 :: Only",
-        "License :: OSI Approved :: MIT License",
-        "Operating System :: OS Independent",
-    ],
-    python_requires=">=3.5",
-)
+#!/usr/bin/env python
+import platform
+
+import setuptools
+
+if (
+    platform.processor() == "arm" or platform.processor() == "i386"
+) and platform.system() == "Darwin":
+    tensorflow_os = "tensorflow-macos>=2.10.0,<=2.15.0"
+else:
+    tensorflow_os = "tensorflow>=2.12.0,<=2.15.0"
+
+with open("README.md", "r") as fh:
+    long_description = fh.read()
+
+setuptools.setup(
+    name="decimer",
+    version="2.7.0",
+    author="Kohulan Rajan",
+    author_email="kohulan.rajan@uni-jena.de",
+    maintainer="Kohulan Rajan, Otto Brinkhaus ",
+    maintainer_email="kohulan.rajan@uni-jena.de, otto.brinkhaus@uni-jena.de",
+    description="DECIMER 2.6.0: Deep Learning for Chemical Image Recognition using Efficient-Net V2 + Transformer",
+    long_description=long_description,
+    long_description_content_type="text/markdown",
+    entry_points={
+        "console_scripts": ["decimer = DECIMER.decimer:main"],
+    },
+    url="https://github.com/Kohulan/DECIMER-Image_Transformer",
+    packages=setuptools.find_packages(),
+    license="MIT",
+    install_requires=[
+        tensorflow_os,
+        "opencv-python",
+        "pystow",
+        "pillow-heif",
+        "efficientnet",
+        "selfies",
+        "pyyaml",
+    ],
+    package_data={"DECIMER": ["repack/*.*", "efficientnetv2/*.*", "Utils/*.*"]},
+    classifiers=[
+        "Programming Language :: Python :: 3",
+        "Programming Language :: Python :: 3.5",
+        "Programming Language :: Python :: 3.6",
+        "Programming Language :: Python :: 3.7",
+        "Programming Language :: Python :: 3.8",
+        "Programming Language :: Python :: 3.9",
+        "Programming Language :: Python :: 3.10",
+        "Programming Language :: Python :: 3 :: Only",
+        "License :: OSI Approved :: MIT License",
+        "Operating System :: OS Independent",
+    ],
+    python_requires=">=3.5",
+)

decimer-2.5.0/DECIMER/decimer.py

@@ -1,162 +0,0 @@
-import logging
-import os
-import pickle
-import sys
-from typing import List
-from typing import Tuple
-
-import pystow
-import tensorflow as tf
-
-import DECIMER.config as config
-import DECIMER.utils as utils
-
-# Silence tensorflow model loading warnings.
-logging.getLogger("absl").setLevel("ERROR")
-
-# Silence tensorflow errors - not recommended if your model is not working properly.
-os.environ["TF_CPP_MIN_LOG_LEVEL"] = "3"
-
-# Set the absolute path
-HERE = os.path.dirname(os.path.abspath(__file__))
-
-# Set model to run on default GPU and allow memory to grow as much as needed.
-# This allows us to run multiple instances of inference in the same GPU.
-os.environ["CUDA_VISIBLE_DEVICES"] = "0"
-gpus = tf.config.experimental.list_physical_devices("GPU")
-for gpu in gpus:
-    tf.config.experimental.set_memory_growth(gpu, True)
-
-# Set path
-default_path = pystow.join("DECIMER-V2")
-
-# download models to a default location
-utils.ensure_model(default_path=default_path)
-
-# Load important pickle files which consists the tokenizers and the maxlength setting
-tokenizer = pickle.load(
-    open(
-        os.path.join(
-            default_path.as_posix(), "DECIMER_model", "assets", "tokenizer_SMILES.pkl"
-        ),
-        "rb",
-    )
-)
-
-
-def main():
-    """
-    This function take the path of the image as user input
-    and returns the predicted SMILES as output in CLI.
-
-    Agrs:
-        str: image_path
-
-    Returns:
-        str: predicted SMILES
-
-    """
-    if len(sys.argv) != 2:
-        print("Usage: {} $image_path".format(sys.argv[0]))
-    else:
-        SMILES = predict_SMILES(sys.argv[1])
-        print(SMILES)
-
-
-def detokenize_output(predicted_array: int) -> str:
-    """
-    This function takes the predited tokens from the DECIMER model
-    and returns the decoded SMILES string.
-
-    Args:
-        predicted_array (int): Predicted tokens from DECIMER
-
-    Returns:
-        (str): SMILES representation of the molecule
-    """
-    outputs = [tokenizer.index_word[i] for i in predicted_array[0].numpy()]
-    prediction = (
-        "".join([str(elem) for elem in outputs])
-        .replace("<start>", "")
-        .replace("<end>", "")
-    )
-    return prediction
-
-
-def detokenize_output_add_confidence(
-    predicted_array: tf.Tensor,
-    confidence_array: tf.Tensor,
-) -> List[Tuple[str, float]]:
-    """
-    This function takes the predicted array of tokens as well as the confidence values
-    returned by the Transformer Decoder and returns a list of tuples
-    that contain each token of the predicted SMILES string and the confidence
-    value.
-
-    Args:
-        predicted_array (tf.Tensor): Transformer Decoder output array (predicted tokens)
-
-    Returns:
-        str: SMILES string
-    """
-    prediction_with_confidence = [
-        (
-            tokenizer.index_word[predicted_array[0].numpy()[i]],
-            confidence_array[i].numpy(),
-        )
-        for i in range(len(confidence_array))
-    ]
-    # remove start and end tokens
-    prediction_with_confidence_ = prediction_with_confidence[1:-1]
-
-    decoded_prediction_with_confidence = list(
-        [(utils.decoder(tok), conf) for tok, conf in prediction_with_confidence_]
-    )
-    decoded_prediction_with_confidence.append(prediction_with_confidence_[-1])
-    return decoded_prediction_with_confidence
-
-
-# Load DECIMER model_packed
-DECIMER_V2 = tf.saved_model.load(default_path.as_posix() + "/DECIMER_model/")
-
-
-def predict_SMILES(image_path: str) -> str:
-    """
-    This function takes an image path (str) and returns the SMILES
-    representation of the depicted molecule (str).
-
-    Args:
-        image_path (str): Path of chemical structure depiction image
-
-    Returns:
-        (str): SMILES representation of the molecule in the input image
-    """
-    chemical_structure = config.decode_image(image_path)
-    predicted_tokens, _ = DECIMER_V2(chemical_structure)
-    predicted_SMILES = utils.decoder(detokenize_output(predicted_tokens))
-    return predicted_SMILES
-
-
-def predict_SMILES_with_confidence(image_path: str) -> List[Tuple[str, float]]:
-    """
-    This function takes an image path (str) and returns a list of tuples
-    that contain each token of the predicted SMILES string and the confidence
-    level from the last layer of the Transformer decoder.
-
-    Args:
-        image_path (str): Path of chemical structure depiction image
-
-    Returns:
-        (List[Tuple[str, float]]): Tuples that contain the tokens and the confidence
-            values of the predicted SMILES
-    """
-    decodedImage = config.decode_image(image_path)
-    predicted_tokens, confidence_values = DECIMER_V2(tf.constant(decodedImage))
-    predicted_SMILES_with_confidence = detokenize_output_add_confidence(
-        predicted_tokens, confidence_values
-    )
-    return predicted_SMILES_with_confidence
-
-
-if __name__ == "__main__":
-    main()