FoodVision Big initial

FoodVision Big - Food 101 Dataset을 구분하는 모델

09_pretrained_effentb1_feature_extractor_food_101_20_percent.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0c3241fc09d7152d840c43e94fed68cf355c52e609d4408dcb89ea8cb40f8251
+size 27031802

app.py

@@ -0,0 +1,64 @@
+###1. 라이브러리, 클래스 이름 불러오기
+import gradio as gr
+import os
+import torch
+
+from model import create_effnetb1_model
+from timeit import default_timer as timer
+from typing import Tuple, Dict
+
+with open("class_names.txt", "r") as f:
+  class_names = [food_name.strip() for food_name in f.readlines()]
+
+### 2. Model, transforms 준비
+effnetb1, effnetb1_transforms = create_effnetb1_model(num_classes=101)
+
+#load saved weights
+effnetb1.load_state_dict(
+    torch.load(f="09_pretrained_effentb1_feature_extractor_food_101_20_percent.pth",
+               map_location=torch.device("cpu"))
+)
+
+### 3. Predict functions
+def predict(img) -> Tuple[Dict, float]:
+  #timer 시작
+  start_time = timer()
+  #image effnetb1 입력형태로 변환
+  img = effnetb1_transforms(img).unsqueeze(0) #batch dimension 0번째 차원에 더하기
+  #예측하기
+  effnetb1.eval()
+  with torch.inference_mode():
+    #image -> prediction logits -> prediction probability
+    pred_probs = torch.softmax(effnetb1(img), dim=1)
+
+  #prediction label, pred probability dictionary
+  pred_labels_and_probs = {class_names[i]: float(pred_probs[0][i]) for i in range(len(class_names))}
+  #예측 시간 계산
+  end_time = timer()
+  pred_time = round(end_time-start_time, 4)
+  #pred dict, pred time 돌려주기
+  return pred_labels_and_probs, pred_time
+
+### 4. Gradio app ###
+
+title = "FoodVision Big"
+description = "An EfficientNetB1 feature extractor for 101 classes of food "
+article = "Created at 09. PyTorch Model Deployment."
+
+#example list를 demo app 내부 경로로 수정
+
+#예시 그림의 파일 경로 가져오기!
+example_list = [["examples/" + example] for example in os.listdir("examples")]
+example_list
+
+#Gradio demo
+demo  = gr.Interface(fn=predict,
+                     inputs=gr.Image(type="pil"),
+                     outputs=[gr.Label(num_top_classes=3, label="Predictions"),
+                              gr.Number(label="Prediction time (s)")],
+                     examples= example_list,
+                     title=title,
+                     description=description,
+                     article=article)
+#데모 실행
+demo.launch(debug=False) #디버그 방식 끄기

class_names.txt

@@ -0,0 +1,101 @@
+apple_pie
+baby_back_ribs
+baklava
+beef_carpaccio
+beef_tartare
+beet_salad
+beignets
+bibimbap
+bread_pudding
+breakfast_burrito
+bruschetta
+caesar_salad
+cannoli
+caprese_salad
+carrot_cake
+ceviche
+cheese_plate
+cheesecake
+chicken_curry
+chicken_quesadilla
+chicken_wings
+chocolate_cake
+chocolate_mousse
+churros
+clam_chowder
+club_sandwich
+crab_cakes
+creme_brulee
+croque_madame
+cup_cakes
+deviled_eggs
+donuts
+dumplings
+edamame
+eggs_benedict
+escargots
+falafel
+filet_mignon
+fish_and_chips
+foie_gras
+french_fries
+french_onion_soup
+french_toast
+fried_calamari
+fried_rice
+frozen_yogurt
+garlic_bread
+gnocchi
+greek_salad
+grilled_cheese_sandwich
+grilled_salmon
+guacamole
+gyoza
+hamburger
+hot_and_sour_soup
+hot_dog
+huevos_rancheros
+hummus
+ice_cream
+lasagna
+lobster_bisque
+lobster_roll_sandwich
+macaroni_and_cheese
+macarons
+miso_soup
+mussels
+nachos
+omelette
+onion_rings
+oysters
+pad_thai
+paella
+pancakes
+panna_cotta
+peking_duck
+pho
+pizza
+pork_chop
+poutine
+prime_rib
+pulled_pork_sandwich
+ramen
+ravioli
+red_velvet_cake
+risotto
+samosa
+sashimi
+scallops
+seaweed_salad
+shrimp_and_grits
+spaghetti_bolognese
+spaghetti_carbonara
+spring_rolls
+steak
+strawberry_shortcake
+sushi
+tacos
+takoyaki
+tiramisu
+tuna_tartare
+waffles

model.py

@@ -0,0 +1,21 @@
+
+import torch
+import torchvision
+
+from torch import nn
+
+def create_effnetb1_model(num_classes:int=101):
+  # 1,2,3 EffNetB1의 미리 훈련된 weights, transforms, model 얻기
+  weights = torchvision.models.EfficientNet_B1_Weights.DEFAULT
+  transforms = weights.transforms()
+  model = torchvision.models.efficientnet_b1(weights=weights)
+
+  # 4. 기본 모델의 layer 고정
+  for param in model.parameters():
+    param.requires_grad = False
+  # 5. classifier head 바꾸기
+  model.classifier = nn.Sequential(
+      nn.Dropout(p=0.3, inplace=True),
+      nn.Linear(in_features=1280, out_features=num_classes)
+  )
+  return model, transforms

requirements.txt

@@ -0,0 +1,3 @@
+torch==2.1.0
+torchvision==0.16.0
+gr==1.23.2