Merge pull request #14 from hardikdava/main

Dashboard

README.md

@@ -177,3 +177,11 @@ Notes:
 * For all practical purposes the ids are same as row number and is reset after every addition or removal
 
 
+
+## Dashboard:
+
+Start dashboard application using following command.
+```
+streamlit run frontend/dashboard.py --server.port 8000
+```
+

frontend/dashboard.py

@@ -0,0 +1,49 @@
+import streamlit as st
+from yoloexplorer import Explorer
+from streamlit_image_select import image_select
+from typing import List
+
+st.title("YOLOExplorer")
+
+with st.sidebar:
+    default_data_yaml = "coco128.yaml"
+    st.text(f"Dataset: {default_data_yaml}")
+    if default_data_yaml:
+        if 'data_yaml' not in st.session_state:
+            st.session_state['data_yaml'] = default_data_yaml
+        if 'session' not in st.session_state:
+            st.session_state['session'] = Explorer(default_data_yaml, model=None)
+            dataset_info = st.session_state['session'].dataset_info
+
+            st.write("**Dataset Information :**")
+
+            for dataset_type in ["train", "val", "tets"]:
+                st.text(f"{dataset_type.capitalize()} paths :")
+                if isinstance(dataset_info['train'], str):
+                    st.text(dataset_info['train'])
+                elif isinstance(dataset_info['train', List]):
+                    for p in dataset_info['train']:
+                        st.text(p)
+
+st.write("**Dataset Visualization**")
+
+# TODO: Only show one of the sample dataset at same time
+img_train_checkbox_w, img_val_checkbox_w, img_test_checkbox_w, num_sample_w = st.columns(4)
+with img_train_checkbox_w:
+    img_train_checkbox = st.checkbox(label="Train")
+with img_val_checkbox_w:
+    img_val_checkbox = st.checkbox(label="Val")
+with img_test_checkbox_w:
+    img_test_checkbox = st.checkbox(label="Test")
+with num_sample_w:
+    num_samples_show = st.number_input(label="Numbers of Images", value=10)
+
+
+if num_samples_show:
+    if img_train_checkbox:
+        img_url = "/home/hd/zidane.jpg"  # Put path to your image
+        train_image_paths = [img_url] * int(num_samples_show)
+        # st.image(train_image_paths, use_column_width="always")
+        clicked = image_select("Training Samples", images=train_image_paths)
+
+

requirements.txt

@@ -1,4 +1,6 @@
 lancedb
 duckdb
 scikit-learn
-ultralytics@git+https://github.com/ultralytics/ultralytics.git@embeddings
+ultralytics@git+https://github.com/ultralytics/ultralytics.git@embeddings
+# streamlit # To run dashboard
+# streamlit-image-select # For image gallery