style: format code with Autopep8, Black and isort

configs/config.py

@@ -27,8 +27,10 @@
     "num_classes": 5270,
     # 'input_shape': (180,180,3)
     "train_from_scratch": 0,  # 0 Means Use pretrained weights 1 Means train from scratch
-    "layer_count_trainable": 20,  # Finetunable layer count if pretrained weights are used.,
-    "model_name": "inception-v3",  # Currently supported = "inception-v3, resnet-101, efficientnet-b7, exception_min"
+    # Finetunable layer count if pretrained weights are used.,
+    "layer_count_trainable": 20,
+    # Currently supported = "inception-v3, resnet-101, efficientnet-b7, exception_min"
+    "model_name": "inception-v3",
     "INIT_LR": 1e-4,
     "MAX_LR": 1e-3,
     "base_learning_rate": 1e-3,
@@ -48,9 +50,11 @@
     "es_mode": "auto",
     "es_restore_best_weights": False,
     "optimizer": "adam",  # currently only adam is available
-    "learning_rate_decay_strategy": "polynomial",  # available  = polynomial, exponential, cyclic, no(if given no there wont be any strategy training will be on the base learning rate)
+    # available  = polynomial, exponential, cyclic, no(if given no there wont be any strategy training will be on the base learning rate)
+    "learning_rate_decay_strategy": "polynomial",
     "reduce_lr_init": 0,  # to use reduce lr or not if set to zero not used -> Train Callback
-    "loss": "categorical_crossentropy",  # Focal loss can be also be used for focal loss please mention focalloss
+    # Focal loss can be also be used for focal loss please mention focalloss
+    "loss": "categorical_crossentropy",
     "earlystopping_init": 0,  # Whether to use Early stopping or not if set to 0 not used
     "save_checkpoint_per_epoch": 0,
     "save_best_checkpoint": 1,

infer_class.py

@@ -1,7 +1,9 @@
 import tensorflow as tf
 from tensorflow import keras
+
+from src.metrics import f1_m, precision_m, recall_m
 from src.model import ModelMaker
-from src.metrics import precision_m, recall_m, f1_m
+
 
 # Class method implementation for Multiple Deployment strategies
 class InferModel:

inference_benchmark.py

@@ -1,14 +1,14 @@
-import tensorflow as tf 
-from tensorflow import keras
-from infer_class import InferModel
-import numpy as np 
 import time
 
+import numpy as np
+import tensorflow as tf
+from tensorflow import keras
+
+from infer_class import InferModel
 
 model_path = "../cdiscount_challenge_code_local/models/model-best.h5"
 
 infer_model = InferModel(model_path, "inceptionv3")
 
 
 model = infer_model.direct_model_load()
-

model_api.py

@@ -1,28 +1,28 @@
 # Class method implementation for Multiple Deployment str
-import tensorflow as tf
-from tensorflow import keras
-from src.model import ModelMaker
-from infer_class import InferModel
+import argparse
+import asyncio
+import os
+import time
+from contextlib import contextmanager
+from pathlib import Path
+from tempfile import TemporaryDirectory
+
+import cv2
 import fastapi
-from fastapi import FastAPI, Query
 import numpy as np
+import pandas as pd
+import tensorflow as tf
 import uvicorn
-import cv2
-from starlette.middleware.cors import CORSMiddleware
-from contextlib import contextmanager
-import asyncio
-from tempfile import TemporaryDirectory
-from pathlib import Path
+from fastapi import FastAPI, Query
+from keras.preprocessing.image import ImageDataGenerator, img_to_array, load_img
 from PIL import Image
-from src.dataprocessor import BSONIterator, CDiscountProcessor
-from keras.preprocessing.image import ImageDataGenerator
-from keras.preprocessing.image import load_img, img_to_array
-import argparse
-import time
-import pandas as pd
-from configs import config
+from starlette.middleware.cors import CORSMiddleware
+from tensorflow import keras
 
-import os
+from configs import config
+from infer_class import InferModel
+from src.dataprocessor import BSONIterator, CDiscountProcessor
+from src.model import ModelMaker
 
 os.environ["CUDA_DEVICE_ORDER"] = "PCI_BUS_ID"  # see issue #152
 os.environ["CUDA_VISIBLE_DEVICES"] = ""
@@ -67,7 +67,8 @@ def predict_class(file: fastapi.UploadFile = fastapi.File(...)):
         print(filepath)
         image = cv2.imread(str(filepath))
         print("Image shape before preprocessing:{}".format(image.shape))
-        image_resized = cv2.resize(image, (180, 180), interpolation=cv2.INTER_AREA)
+        image_resized = cv2.resize(
+            image, (180, 180), interpolation=cv2.INTER_AREA)
 
         x = img_to_array(image_resized)
         x = ImageDataGenerator().random_transform(x)

model_converter.py

@@ -1,7 +1,8 @@
 import tensorflow as tf
 from tensorflow import keras
+
+from src.metrics import f1_m, precision_m, recall_m
 from src.model import ModelMaker
-from src.metrics import precision_m, recall_m, f1_m
 
 """
 

src/callbacks.py

@@ -1,13 +1,15 @@
 # Class to define different callbacks and learning rate strategies
-import tensorflow as tf
-import wandb
-from configs import config
-import tensorflow_addons as tfa
-from datetime import datetime
 import os
 import shutil
+from datetime import datetime
+
+import tensorflow as tf
+import tensorflow_addons as tfa
+import wandb
 from wandb.keras import WandbCallback
 
+from configs import config
+
 
 # Custom class to log learning rate to Wandb
 class LRLogger(tf.keras.callbacks.Callback):
@@ -107,7 +109,8 @@ def set_callbacks_and_optimizer(self):
         # Save best checkpoint based on metric monitoring
 
         filepath = (
-            checkpoints_path + "/" + "weights.best_{epoch:02d}-{val_accuracy:.2f}.hdf5"
+            checkpoints_path + "/" +
+            "weights.best_{epoch:02d}-{val_accuracy:.2f}.hdf5"
         )
         checkpoint_best = tf.keras.callbacks.ModelCheckpoint(
             filepath,

src/dataprocessor.py

@@ -1,28 +1,33 @@
-import os, sys, math, io
-import numpy as np
-import pandas as pd
+import io
+import math
 import multiprocessing as mp
-import bson
+import os
 import struct
+import sys
+import threading
+import time
+from collections import defaultdict
 
-
+import bson
 import matplotlib.pyplot as plt
+import numpy as np
+import pandas as pd
+import pymongo
 import skimage.io
-from keras.preprocessing.image import load_img, img_to_array
-from configs import config
-from tqdm import tqdm
-from keras.preprocessing.image import Iterator
-from keras.preprocessing.image import ImageDataGenerator
+import skimage.io as skio
 from keras import backend as K
+from keras.preprocessing.image import (
+    ImageDataGenerator,
+    Iterator,
+    img_to_array,
+    load_img,
+)
 from PIL import Image
 from pymongo import MongoClient
-import io
-import pymongo
-import skimage.io as skio
-import threading
-import time
-from collections import defaultdict
 from tqdm import *
+from tqdm import tqdm
+
+from configs import config
 
 
 class CDiscountProcessor:
@@ -60,7 +65,6 @@ def make_category_tables(self, categories_df):
         idx2cat = {}
         i = 0
         for ir in categories_df.itertuples():
-
             category_id = ir[0]
             category_idx = ir[4]
             cat2idx[category_id] = category_idx
@@ -130,7 +134,8 @@ def make_val_set(self, df, split_percentage=0.2, drop_percentage=0.0):
                 # Randomly choose the products that become part of the validation set.
                 val_size = int(len(product_ids) * split_percentage)
                 if val_size > 0:
-                    val_ids = np.random.choice(product_ids, val_size, replace=False)
+                    val_ids = np.random.choice(
+                        product_ids, val_size, replace=False)
                 else:
                     val_ids = []
 
@@ -150,16 +155,17 @@ def make_val_set(self, df, split_percentage=0.2, drop_percentage=0.0):
         return train_df, val_df
 
     def generate_lookup_table(self):
-        self.categories_df = pd.read_csv(self.category_path, index_col="category_id")
+        self.categories_df = pd.read_csv(
+            self.category_path, index_col="category_id")
         self.categories_df["category_idx"] = pd.Series(
             range(len(self.categories_df)), index=self.categories_df.index
         )
-        self.cat2idx, self.idx2cat = self.make_category_tables(self.categories_df)
+        self.cat2idx, self.idx2cat = self.make_category_tables(
+            self.categories_df)
         # Testing
         print(self.cat2idx[1000012755], self.idx2cat[4])
 
     def read_images_load_train_val(self):
-
         self.train_offsets_df = self.read_bson(
             self.train_path, num_records=self.num_train_products, with_categories=True
         )
@@ -258,7 +264,6 @@ def __init__(
         shuffle=False,
         seed=None,
     ):
-
         self.file = bson_file
         self.images_df = images_df
         self.offsets_df = offsets_df
@@ -270,16 +275,20 @@ def __init__(
         self.image_shape = self.target_size + (3,)
 
         print(
-            "Found %d images belonging to %d classes." % (self.samples, self.num_class)
+            "Found %d images belonging to %d classes." % (
+                self.samples, self.num_class)
         )
 
-        super(BSONIterator, self).__init__(self.samples, batch_size, shuffle, seed)
+        super(BSONIterator, self).__init__(
+            self.samples, batch_size, shuffle, seed)
         self.lock = lock
 
     def _get_batches_of_transformed_samples(self, index_array):
-        batch_x = np.zeros((len(index_array),) + self.image_shape, dtype=K.floatx())
+        batch_x = np.zeros((len(index_array),) +
+                           self.image_shape, dtype=K.floatx())
         if self.with_labels:
-            batch_y = np.zeros((len(batch_x), self.num_class), dtype=K.floatx())
+            batch_y = np.zeros(
+                (len(batch_x), self.num_class), dtype=K.floatx())
 
         for i, j in enumerate(index_array):
             # Protect file and dataframe access with a lock.

src/model.py

@@ -1,7 +1,8 @@
+import tensorflow
 import tensorflow as tf
 from tensorflow import keras
 from tensorflow.keras import layers
-import tensorflow
+
 from configs import config
 
 
@@ -11,7 +12,8 @@ def __init__(self):
         self.input_shape = config.model_parameter_dict["target_size"] + (3,)
         print("==" * 10)
         print(
-            "The default input shape of the model will be {}".format(self.input_shape)
+            "The default input shape of the model will be {}".format(
+                self.input_shape)
         )
         self.num_classes = config.model_parameter_dict["num_classes"]
         print(

train.py

@@ -1,14 +1,15 @@
-from src.callbacks import CallbacksFormatter
-from src.dataprocessor import CDiscountProcessor
-from src.metrics import precision_m, recall_m, f1_m
-from src.model import ModelMaker
-from configs import config
-import tensorflow as tf
-from tensorflow import keras
 import argparse
+
+import tensorflow as tf
 import wandb
+from tensorflow import keras
 from wandb.keras import WandbCallback
 
+from configs import config
+from src.callbacks import CallbacksFormatter
+from src.dataprocessor import CDiscountProcessor
+from src.metrics import f1_m, precision_m, recall_m
+from src.model import ModelMaker
 
 cdiscount_processor = CDiscountProcessor()
 modelmaker = ModelMaker()

web_app.py

@@ -29,7 +29,8 @@
     if image is not None:
         files = {"file": image.getvalue()}
         with st.spinner("API Request Initiated.Waiting for Response"):
-            res = requests.post(("http://localhost:5001/predict/"), files=files)
+            res = requests.post(
+                ("http://localhost:5001/predict/"), files=files)
         st.success("Request completed Successfully")
         text = res.json()
         st.session_state.key = text
@@ -52,4 +53,5 @@
 
     col4, col5 = st.columns(2)
     col4.metric(label="Response Time", value=text["response_time"])
-    col5.metric(label="Model Inference time", value=text["model_inference_time"])
+    col5.metric(label="Model Inference time",
+                value=text["model_inference_time"])