Merge pull request #363 from JdeRobot/issue-362

Upgrade streamlit and increase model file limit size

Author: dpascualhe

app.py

@@ -1,5 +1,4 @@
 import streamlit as st
-import os
 import sys
 import subprocess
 from tabs.dataset_viewer import dataset_viewer_tab
@@ -67,6 +66,10 @@ def browse_folder():
         return None
 
 
+def browse_dataset_path():
+    st.session_state.dataset_path = browse_folder()
+
+
 st.set_page_config(page_title="PerceptionMetrics", layout="wide")
 
 PAGES = {
@@ -77,8 +80,8 @@ def browse_folder():
 
 # Initialize commonly used session state keys
 st.session_state.setdefault("dataset_path", "")
-st.session_state.setdefault("dataset_type_selectbox", "COCO")
-st.session_state.setdefault("split_selectbox", "val")
+st.session_state.setdefault("dataset_type", "COCO")
+st.session_state.setdefault("split", "val")
 st.session_state.setdefault("config_option", "Manual Configuration")
 st.session_state.setdefault("confidence_threshold", 0.5)
 st.session_state.setdefault("nms_threshold", 0.5)
@@ -95,66 +98,44 @@ def browse_folder():
         # First row: Type and Split
         col1, col2 = st.columns(2)
         with col1:
-            dataset_type_selectbox = st.selectbox(
+            st.selectbox(
                 "Type",
                 ["COCO", "YOLO"],
-                key="dataset_type_selectbox",
+                key="dataset_type",
             )
         with col2:
             st.selectbox(
                 "Split",
                 ["train", "val", "test"],
-                key="split_selectbox",
+                key="split",
             )
 
         # Second row: Path and Browse button
         col1, col2 = st.columns([3, 1])
         with col1:
-            dataset_path_input = st.text_input(
-                "Dataset Folder",
-                value=st.session_state.get("dataset_path", ""),
-                key="dataset_path_input",
-            )
+            st.text_input("Dataset Folder", key="dataset_path")
         with col2:
             st.markdown(
                 "<div style='margin-bottom: 1.75rem;'></div>", unsafe_allow_html=True
             )
-            if st.button("Browse", key="browse_button"):
-                folder = browse_folder()
-                if folder and os.path.isdir(folder):
-                    st.session_state["dataset_path"] = folder
-                    st.rerun()
-                elif folder is not None:
-                    st.warning("Selected path is not a valid folder.")
-                else:
-                    st.warning(
-                        "Could not open folder browser. Please enter the path manually"
-                    )
-
-        if dataset_path_input != st.session_state.get("dataset_path", ""):
-            st.session_state["dataset_path"] = dataset_path_input
-        if dataset_type_selectbox != st.session_state.get("dataset_type", ""):
-            st.session_state["dataset_type"] = dataset_type_selectbox
+            st.button("Browse", on_click=browse_dataset_path)
 
         # Additional input for YOLO config file
-        if dataset_type_selectbox == "YOLO":
-            dataset_config_file_uploader = st.file_uploader(
+        if st.session_state.get("dataset_type", "COCO") == "YOLO":
+            st.file_uploader(
                 "Dataset Configuration (.yaml)",
                 type=["yaml"],
                 key="dataset_config_file",
                 help="Upload a YAML dataset configuration file.",
             )
-            if dataset_config_file_uploader != st.session_state.get(
-                "dataset_config_file", None
-            ):
-                st.session_state["dataset_config_file"] = dataset_config_file_uploader
 
     with st.expander("Model Inputs", expanded=False):
         st.file_uploader(
             "Model File (.pt, .onnx, .h5, .pb, .pth, .torchscript)",
             type=["pt", "onnx", "h5", "pb", "pth", "torchscript"],
             key="model_file",
             help="Upload your trained model file.",
+            max_upload_size=1024,  # MB
         )
         st.file_uploader(
             "Ontology File (.json)",
@@ -254,14 +235,15 @@ def browse_folder():
                     key="resize_width",
                     help="Width to resize images for inference",
                 )
+
         # Load model action in sidebar
         from perceptionmetrics.models.torch_detection import TorchImageDetectionModel
         import json, tempfile
 
         load_model_btn = st.button(
             "Load Model",
             type="primary",
-            use_container_width=True,
+            width="stretch",
             help="Load and save the model for use in the Inference tab",
             key="sidebar_load_model_btn",
         )

pyproject.toml

@@ -24,7 +24,7 @@ click = "^8.1.8"
 tensorboard = "^2.18.0"
 pycocotools = { version = "^2.0.7", markers = "sys_platform != 'win32'" }
 pycocotools-windows = { version = "^2.0.0.2", markers = "sys_platform == 'win32'" }
-Streamlit = "1.46.0"
+Streamlit = "1.54.0"
 streamlit-image-select = "^0.6.0"
 supervision = "^0.18.0"
 

tabs/dataset_viewer.py

@@ -7,7 +7,6 @@ def dataset_viewer_tab():
     import tempfile
     from perceptionmetrics.datasets.coco import CocoDataset
     from perceptionmetrics.datasets.yolo import YOLODataset
-    import supervision as sv
     import numpy as np
     from PIL import Image
     from supervision.draw.color import ColorPalette
@@ -16,8 +15,8 @@ def dataset_viewer_tab():
 
     # Get inputs from session state
     dataset_path = st.session_state.get("dataset_path", "")
-    dataset_type = st.session_state.get("dataset_type_selectbox", "COCO").lower()
-    split = st.session_state.get("split_selectbox", "val")
+    dataset_type = st.session_state.get("dataset_type", "COCO").lower()
+    split = st.session_state.get("split", "val")
 
     # Header row only
     st.header("Dataset Viewer")
@@ -27,7 +26,6 @@ def dataset_viewer_tab():
         return
 
     # Setup paths and pagination
-           # Setup paths and pagination
     if dataset_type == "coco":
         img_dir = os.path.join(dataset_path, f"images/{split}2017")
         ann_file = os.path.join(
@@ -162,7 +160,7 @@ def dataset_viewer_tab():
 
     # Pagination
     IMAGES_PER_PAGE = 12
-    total_images, total_pages = (
+    _, total_pages = (
         len(image_files),
         (len(image_files) + IMAGES_PER_PAGE - 1) // IMAGES_PER_PAGE,
     )
@@ -197,7 +195,7 @@ def dataset_viewer_tab():
             st.rerun()
     with col2:
         st.markdown(
-            f"<div style='text-align:center;font-weight:bold;'>Page {current_page+1} of {total_pages}</div>",
+            f"<div style='text-align:center;font-weight:bold;'>Page {current_page + 1} of {total_pages}</div>",
             unsafe_allow_html=True,
         )
     with col3:
@@ -220,7 +218,7 @@ def dataset_viewer_tab():
         col1, col2, col3 = st.columns([4, 1, 1])
         with col1:
             selected_img = st.selectbox(
-                "Search image:", options=image_files, key="search_image_selectbox"
+                "Search image:", options=image_files, key="search_image"
             )
         with col2:
             st.markdown(
@@ -231,7 +229,7 @@ def dataset_viewer_tab():
                 st.session_state[page_key] = new_page
                 st.session_state[
                     f"img_select_all_{dataset_path}_{split}_{new_page}"
-                ] = (image_files.index(selected_img) % IMAGES_PER_PAGE)
+                ] = image_files.index(selected_img) % IMAGES_PER_PAGE
                 st.session_state["show_search_dropdown"] = False
                 st.rerun()
         with col3:
@@ -252,7 +250,7 @@ def dataset_viewer_tab():
             label="",
             images=image_paths,
             captions=sample_images,
-            use_container_width=True,
+            use_container_width=False,
             key=img_select_key,
             index=img_select_index,
         )
@@ -314,7 +312,7 @@ def dataset_viewer_tab():
                 except AttributeError:
                     resample = Image.LANCZOS
                 annotated_pil.thumbnail((500, 500), resample)
-                st.image(annotated_pil, use_container_width=False)
+                st.image(annotated_pil, width="content")
             else:
                 st.warning("No annotation found for this image.")
         except Exception as e:

tabs/evaluator.py

@@ -2,8 +2,6 @@
 import os
 import tempfile
 import json
-import pandas as pd
-from perceptionmetrics.models.torch_detection import TorchImageDetectionModel
 from perceptionmetrics.datasets.coco import CocoDataset
 
 
@@ -19,8 +17,8 @@ def evaluator_tab():
 
     # Check for dataset from sidebar inputs
     dataset_path = st.session_state.get("dataset_path", "")
-    dataset_type = st.session_state.get("dataset_type_selectbox", "Coco")
-    split = st.session_state.get("split_selectbox", "val")
+    dataset_type = st.session_state.get("dataset_type", "Coco")
+    split = st.session_state.get("split", "val")
 
     # Try to get existing dataset from session state first
     dataset_key = f"{dataset_path}_{split}"
@@ -134,11 +132,6 @@ def evaluator_tab():
                 if save_predictions:
                     predictions_outdir = tempfile.mkdtemp(prefix="eval_predictions_")
 
-                # Use model config as is (no confidence threshold override)
-                eval_config = model.model_cfg.copy()
-
-                # Ready to evaluate
-
                 # Create progress bar for evaluation
                 progress_bar = st.progress(0)
                 status_text = st.empty()
@@ -205,7 +198,7 @@ def metrics_callback(metrics_df, processed, total):
 
                         with intermediate_table_placeholder.container():
                             st.markdown("#### Per-Class Metrics (Intermediate)")
-                            st.dataframe(display_df, use_container_width=True)
+                            st.dataframe(display_df, width="stretch")
 
                     except Exception as e:
                         st.error(f"Metrics callback error: {e}")
@@ -321,7 +314,7 @@ def display_evaluation_results(results):
         if col in display_df.columns:
             display_df[col] = display_df[col].round(3)
 
-    st.dataframe(display_df, use_container_width=True)
+    st.dataframe(display_df, width="stretch")
 
     # Now display Precision-Recall Curve
     if metrics_factory is not None:
@@ -334,7 +327,6 @@ def display_evaluation_results(results):
 
             # Create the plot using streamlit's plotly integration
             import plotly.graph_objects as go
-            from plotly.subplots import make_subplots
 
             # Create the precision-recall curve
             fig = go.Figure()
@@ -379,7 +371,7 @@ def display_evaluation_results(results):
             fig.update_xaxes(showgrid=True, gridwidth=1, gridcolor="lightgray")
             fig.update_yaxes(showgrid=True, gridwidth=1, gridcolor="lightgray")
 
-            st.plotly_chart(fig, use_container_width=True)
+            st.plotly_chart(fig, width="stretch")
 
         except Exception as e:
             st.error(f"Error plotting precision-recall curve: {e}")
@@ -403,7 +395,6 @@ def display_evaluation_results(results):
             else None
         )
         if curve_data is not None:
-            import io
             import pandas as pd
 
             pr_points_df = pd.DataFrame(
@@ -439,7 +430,7 @@ def display_evaluation_results(results):
         st.write("Columns:", metrics_df.columns.tolist())
 
         st.markdown("**Sample Data:**")
-        st.dataframe(metrics_df.head(), use_container_width=True)
+        st.dataframe(metrics_df.head(), width="stretch")
 
         if "evaluation_config" in st.session_state:
             st.markdown("**Evaluation Configuration:**")

tabs/inference.py

@@ -81,9 +81,7 @@ def inference_tab():
                 result_img = draw_detections(image.copy(), predictions, label_map)
 
                 st.markdown("#### Detection Results")
-                st.image(
-                    result_img, caption="Detection Results", use_container_width=True
-                )
+                st.image(result_img, caption="Detection Results", width="stretch")
 
                 # Display detection statistics
                 if (