modified object detection example

data/dior_show/results/dior1_results.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6d8e9ca5b6ebd77da5cf893e062316121e3145bca6afc993e350aec9780b64fd
+size 5936769

data/dior_show/results/dior2_results.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a7f562867db3dc0cc8fe8d49fd87ce2aadae96715e0dabe0a9dd24cc913a1c31
+size 5936769

data/dior_show/results/dior3_results.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:155efde06760dc3b86684e4cfd96209cb528df9a7b7d2696df49863f23d1e4e6
+size 5936769

data/dior_show/results/dior4_results.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:39d60dcf806580a5092f3f4104fc92101e18863524f34fe2bf909ba66e884973
+size 5936769

pages/object_detection.py

@@ -4,14 +4,14 @@ import streamlit as st
 import matplotlib.pyplot as plt
 import pandas as pd
 import numpy as np
-#import altair as alt
 import plotly.express as px
-
+import pickle
+import random
 
 from PIL import Image
 from transformers import YolosFeatureExtractor, YolosForObjectDetection
 from torchvision.transforms import ToTensor, ToPILImage
-#from utils import load_model_huggingface
+from annotated_text import annotated_text
 
 
 st.set_page_config(layout="wide")
@@ -80,8 +80,7 @@ def plot_results(pil_img, prob, boxes):
             ax.add_patch(plt.Rectangle((xmin, ymin), xmax - xmin, ymax - ymin,
                                     fill=False, color=c, linewidth=3))
             ax.text(xmin, ymin, f"{idx_to_text(cl)}", fontsize=10,
-                    bbox=dict(facecolor=c, alpha=0.8))
-            
+                    bbox=dict(facecolor=c, alpha=0.8))            
     plt.axis('off')
     
     plt.savefig("results_od.png", 
@@ -100,18 +99,6 @@ def return_probas(outputs, threshold):
     return probas, keep
 
 
-
-# def visualize_predictions(image, outputs, threshold):
-# # keep only predictions with confidence >= threshold
-#     # convert predicted boxes from [0; 1] to image scales
-#     bboxes_scaled = rescale_bboxes(outputs.pred_boxes[0, keep].cpu(), image.size)
-
-#     # plot results
-#     plot_results(image, probas[keep], bboxes_scaled)
-
-#     return probas[keep]
-
-
 def visualize_probas(probas, threshold, colors):
     label_df = pd.DataFrame({"label":probas.max(-1).indices.detach().numpy(),
                              "proba":probas.max(-1).values.detach().numpy()})
@@ -136,9 +123,10 @@ def visualize_probas(probas, threshold, colors):
                  color="Item", title="Probability scores")
     st.plotly_chart(fig, use_container_width=True)
 
-    # chart = alt.Chart(top_label_df_agg).mark_bar().encode(x="proba", y="label", 
-    #                                                       color=alt.Color('colors:N', scale=None)).interactive()
-    # st.altair_chart(chart)
+
+cats = ['shirt, blouse', 'top, t-shirt, sweatshirt', 'sweater', 'cardigan', 'jacket', 'vest', 'pants', 'shorts', 'skirt', 'coat', 'dress', 'jumpsuit',
+    'cape', 'glasses', 'hat', 'headband, head covering, hair accessory', 'tie', 'glove', 'watch', 'belt', 'leg warmer', 'tights, stockings', 'sock', 'shoe', 'bag, wallet', 'scarf', 'umbrella', 'hood', 'collar',
+    'lapel', 'epaulette', 'sleeve', 'pocket', 'neckline', 'buckle', 'zipper', 'applique', 'bead', 'bow', 'flower', 'fringe', 'ribbon', 'rivet', 'ruffle', 'sequin', 'tassel']
 
 
 
@@ -182,49 +170,75 @@ st.info("""In this use case, we are going to identify and locate different artic
 
 st.markdown("  ")
 
-images_dior = [os.path.join("data/dior_show",url) for url in os.listdir("data/dior_show") if url != "results"]
+images_dior = [os.path.join("data/dior_show/images",url) for url in os.listdir("data/dior_show/images") if url != "results"]
 columns_img = st.columns(4)
 for img, col in zip(images_dior,columns_img):
     with col:
         st.image(img)
 
-
 st.markdown("  ")
 
 
+st.markdown("### About the model 📚")
+st.markdown("""The object detection model was trained specifically to **detect clothing items** on images. <br> 
+            It is able to detect <b>46</b> different types of clothing items.""", unsafe_allow_html=True)
+
+colors = ["#8ef", "#faa", "#afa", "#fea", "#8ef","#afa"]*7 + ["#8ef", "#faa", "#afa", "#fea"]
+
+cats_annotated = [(g,"","#afa") for g in cats]
+annotated_text([cats_annotated])
+
+# st.markdown("""**Here are the 'objects' the model is able to detect**: <br>
+#             'shirt, blouse', 'top, t-shirt, sweatshirt', 'sweater', 'cardigan', 'jacket', 'vest', 'pants', 'shorts', 'skirt', 
+#             'coat', 'dress', 'jumpsuit', 'cape', 'glasses', 'hat', 'headband, head covering, hair accessory', 'tie', 'glove', 'watch', 
+#             'belt', 'leg warmer', 'tights, stockings', 'sock', 'shoe', 'bag, wallet', 'scarf', 'umbrella', 'hood', 'collar', 'lapel', 
+#             'epaulette', 'sleeve', 'pocket', 'neckline', 'buckle', 'zipper', 'applique', 'bead', 'bow', 'flower', 'fringe', 'ribbon', 'rivet', 
+#             'ruffle', 'sequin', 'tassel'""", unsafe_allow_html=True)
+
+st.markdown("Credits: https://huggingface.co/valentinafeve/yolos-fashionpedia")
+st.markdown("")
+st.markdown("")
+
+
+
 ############## SELECT AN IMAGE ###############
 
-st.markdown("#### Select an image 🖼️")
+st.markdown("### Select an image 🖼️")
 #st.markdown("""**Select an image that you wish to run the Object Detection model on.**""") 
 
-
 image_ = None
-select_image_box = st.radio(
-    "**Select the image you wish to run the model on**",
-    ["Choose an existing image", "Load your own image"],
-    index=None,)# #label_visibility="collapsed")
-
-if select_image_box == "Choose an existing image":
-    fashion_images_path = r"data/dior_show"
-    list_images = os.listdir(fashion_images_path)
-    image_ = st.selectbox("", list_images, label_visibility="collapsed")
+fashion_images_path = r"data/dior_show/images"
+list_images = os.listdir(fashion_images_path)
+image_ = st.selectbox("Select the image you wish to run the model on", list_images)
+image_ = os.path.join(fashion_images_path, image_)#label_visibility="collapsed")
+
+# image_ = None
+# select_image_box = st.radio(
+#     "**Select the image you wish to run the model on**",
+#     ["Choose an existing image", "Load your own image"],
+#     index=None,)# #label_visibility="collapsed")
+
+# if select_image_box == "Choose an existing image":
+#     fashion_images_path = r"data/dior_show/images"
+#     list_images = os.listdir(fashion_images_path)
+#     image_ = st.selectbox("", list_images, label_visibility="collapsed")
     
-    if image_ is not None:
-        image_ = os.path.join(fashion_images_path,image_)
-        st.markdown("You've selected the following image:")
-        st.image(image_, width=300)
-
-elif select_image_box == "Load your own image":
-    image_ = st.file_uploader("Load an image here", 
-                                key="OD_dior", type=['jpg','jpeg','png'], label_visibility="collapsed")
+#     if image_ is not None:
+#         image_ = os.path.join(fashion_images_path,image_)
+#         st.markdown("You've selected the following image:")
+#         st.image(image_, width=300)
+
+# elif select_image_box == "Load your own image":
+#     image_ = st.file_uploader("Load an image here", 
+#                                 key="OD_dior", type=['jpg','jpeg','png'], label_visibility="collapsed")
     
-    st.warning("""**Note**: The model tends to perform better with images of people/clothing items facing forward. 
-           Choose this type of image if you want optimal results.""")
-    st.warning("""**Note:** The model was trained to detect clothing items on a single person. 
-               If your image contains more than one person, the model won't detect the items of the other persons.""")
+#     st.warning("""**Note**: The model tends to perform better with images of people/clothing items facing forward. 
+#            Choose this type of image if you want optimal results.""")
+#     st.warning("""**Note:** The model was trained to detect clothing items on a single person. 
+#                If your image contains more than one person, the model won't detect the items of the other persons.""")
 
-    if image_ is not None:
-        st.image(Image.open(image_), width=300)
+#     if image_ is not None:
+#         st.image(Image.open(image_), width=300)
 
 
 st.markdown("  ")
@@ -234,46 +248,41 @@ st.markdown("  ")
 
 ########## SELECT AN ELEMENT TO DETECT ##################
 
-cats = ['shirt, blouse', 'top, t-shirt, sweatshirt', 'sweater', 'cardigan', 'jacket', 'vest', 'pants', 'shorts', 'skirt', 'coat', 'dress', 'jumpsuit',
-    'cape', 'glasses', 'hat', 'headband, head covering, hair accessory', 'tie', 'glove', 'watch', 'belt', 'leg warmer', 'tights, stockings', 'sock', 'shoe', 'bag, wallet', 'scarf', 'umbrella', 'hood', 'collar',
-    'lapel', 'epaulette', 'sleeve', 'pocket', 'neckline', 'buckle', 'zipper', 'applique', 'bead', 'bow', 'flower', 'fringe', 'ribbon', 'rivet', 'ruffle', 'sequin', 'tassel']
 
 dict_cats = dict(zip(np.arange(len(cats)), cats))
 
-st.markdown("#### Choose the elements you want to detect 👉")
+# st.markdown("#### Choose the elements you want to detect 👉")
 
-# Select one or more elements to detect
-container = st.container()
-selected_options = None
-all = st.checkbox("Select all")
+# # Select one or more elements to detect
+# container = st.container()
+# selected_options = None
+# all = st.checkbox("Select all")
 
-if all:
-    selected_options = container.multiselect("**Select one or more items**", cats, cats)
-else:
-    selected_options = container.multiselect("**Select one or more items**", cats)
+# if all:
+#     selected_options = container.multiselect("**Select one or more items**", cats, cats)
+# else:
+#     selected_options = container.multiselect("**Select one or more items**", cats)
 
 #cats = selected_options 
+selected_options = cats
 dict_cats_final = {key:value for (key,value) in dict_cats.items() if value in selected_options}
 
 
-st.markdown("  ")
-st.markdown("  ")
+# st.markdown("  ")
+# st.markdown("  ")
 
 
 
 ############## SELECT A THRESHOLD ###############
 
-st.markdown("#### Define a threshold for predictions 🔎")
-st.markdown("""Object detection models assign to each element detected a **probability score**. <br>
-            This score represents the model's belief in the accuracy of its prediction for a specific object.
-            """, unsafe_allow_html=True)
+st.markdown("### Define a threshold for predictions 🔎")
+st.markdown("""This section allows you to control how confident you want your model to be with its predictions. <br>
+            Objects that are given a lower score than the chosen threshold will be ignored in the final results.""", unsafe_allow_html=True)
 
-st.warning("**Note:** Objects that are assigned a lower score than the chosen threshold will be ignored in the final results.")
+st.markdown(" Below is an example of probability scores given by object detection models for each element detected.")
 
 
-_, col, _ = st.columns([0.2,0.6,0.2])
-with col:
-    st.image("images/probability_od.png", caption="Example of object detection with probability scores")
+st.image("images/probability_od.png", caption="Example with bounding boxes and probability scores given by object detection models")
 
 st.markdown(" ")
 
@@ -284,13 +293,18 @@ st.markdown("**Select a threshold** ")
 
 threshold = st.slider('**Select a threshold**', min_value=0.5, step=0.05, max_value=1.0, value=0.75, label_visibility="collapsed")
 
-if threshold < 0.6:
-    st.error("""**Warning**: Selecting a low threshold (below 0.6) could lead the model to make errors and detect too many objects.""")
+
+# if threshold < 0.6:
+#     st.error("""**Warning**: Selecting a low threshold (below 0.6) could lead the model to make errors and detect too many objects.""")
 
 st.write("You've selected a threshold at", threshold)
 st.markdown("  ")
 
 
+
+pickle_file_path = r"data/dior_show/results"
+
+
 ############# RUN MODEL ################
 
 run_model = st.button("**Run the model**", type="primary")
@@ -299,19 +313,30 @@ if run_model:
     if image_ != None and selected_options != None and threshold!= None:
         with st.spinner('Wait for it...'):
             ## SELECT IMAGE
+            #st.write(image_)
             image = Image.open(image_)
             image = fix_channels(ToTensor()(image))
 
             ## LOAD OBJECT DETECTION MODEL
             FEATURE_EXTRACTOR_PATH = "hustvl/yolos-small"
             MODEL_PATH = "valentinafeve/yolos-fashionpedia"
-            feature_extractor, model = load_model(FEATURE_EXTRACTOR_PATH, MODEL_PATH)
-            # feature_extractor = YolosFeatureExtractor.from_pretrained('hustvl/yolos-small')
-            # model = YolosForObjectDetection.from_pretrained(MODEL)
+            # feature_extractor, model = load_model(FEATURE_EXTRACTOR_PATH, MODEL_PATH)
 
-            # RUN MODEL ON IMAGE
-            inputs = feature_extractor(images=image, return_tensors="pt")
-            outputs = model(**inputs)
+            # # RUN MODEL ON IMAGE
+            # inputs = feature_extractor(images=image, return_tensors="pt")
+            # outputs = model(**inputs)
+            
+            # Save results 
+            # pickle_file_path = r"data/dior_show/results"
+            # image_name = image_.split('\\')[1][:5]
+            # with open(os.path.join(pickle_file_path, f"{image_name}_results.pkl"), 'wb') as file:
+            #     pickle.dump(outputs, file)
+            
+            image_name = image_.split('\\')[1][:5]
+            path_load_pickle = os.path.join(pickle_file_path, f"{image_name}_results.pkl")
+            with open(path_load_pickle, 'rb') as pickle_file:
+                outputs = pickle.load(pickle_file)
+            
             probas, keep = return_probas(outputs, threshold)
 
             st.markdown("#### See the results ☑️")