upload data

app.py

@@ -0,0 +1,104 @@
+import streamlit as st
+import torch,torchvision
+import detectron2
+import pickle
+from detectron2.utils.logger import setup_logger
+logger = setup_logger()
+from detectron2.engine import DefaultPredictor
+from detectron2.projects import point_rend
+from viz_app import plot_single_image  # Correct import
+from PIL import Image
+import numpy as np
+from detectron2.data import MetadataCatalog, DatasetCatalog
+from detectron2.utils.visualizer import Visualizer
+
+# Define configuration parameters
+output_directories = {
+    "2 Classes": "output/pointrend/",
+    "4 Classes": "output/pointrend_4_cls_7k/",
+}
+
+# Function to load the configuration
+def load_configuration(output_dir):
+    with open(output_dir + "cfg.pickle", "rb") as f:
+        return pickle.load(f)
+
+# Function to run instance segmentation
+def run_instance_segmentation(im, predictor, num_classes):
+    
+    outputs = plot_single_image(im, predictor,num_classes)
+
+    masks = outputs['instances'].pred_masks.cpu().numpy()
+    classes = outputs['instances'].pred_classes.cpu().numpy()
+
+    if num_classes == 2:
+        # Create masks for Nest and Inorganic Material
+        nest_masks = masks[classes == 0]
+        inorganic_material_masks = masks[classes == 1]
+                        
+        # Calculate the total pixel area of Nest masks
+        total_nest_area = np.sum(nest_masks)
+        
+        # Calculate the total area of Inorganic Material masks
+        total_inorganic_area = np.sum(inorganic_material_masks)
+
+    elif num_classes == 4:
+            
+        # Create masks for Nest and Inorganic Material
+        plastic_masks = masks[classes == 0] 
+        fishing_net_masks = masks[classes == 1]
+        rope_cloth_masks = masks[classes == 2] 
+        nest_masks = masks[classes == 3]
+                
+        
+        # Calculate the total pixel area of Nest masks
+        total_nest_area = np.sum(nest_masks)
+        
+        # Calculate the total area of Inorganic Material masks
+        plastic_area = np.sum(plastic_masks)
+        fishing_net_area = np.sum(fishing_net_masks)
+        rope_cloth_area = np.sum(rope_cloth_masks)
+        total_inorganic_area = np.sum([plastic_area, fishing_net_area, rope_cloth_area])
+
+    return total_nest_area, total_inorganic_area
+
+def click_instance_segmentation(image, model_selection, predictor):
+    im = np.array(image)[:, :, ::-1]
+
+    if st.button("Run Instance Segmentation"):
+        num_classes = 2 if model_selection == "2 Classes" else 4
+        total_nest_area, total_inorganic_area = run_instance_segmentation(im, predictor, num_classes)
+
+        # Calculate the percentage of inorganic material within the Nest
+        percentage_inorganic_in_nest = (total_inorganic_area / total_nest_area) * 100
+
+        st.write(f"Percentage of Inorganic Material in Nest: {percentage_inorganic_in_nest:.2f}%")
+    
+
+def app():
+    # Create a sidebar to select the model
+    model_selection = st.sidebar.radio("Select Model", list(output_directories.keys()))
+    output_dir = output_directories[model_selection]
+    cfg = load_configuration(output_dir)
+
+    cfg.MODEL.WEIGHTS = output_dir + "model_best.pth"
+    cfg.MODEL.DEVICE = "cpu"
+    cfg.MODEL.ROI_HEADS.SCORE_THRESH_TEST = 0.70
+    predictor = DefaultPredictor(cfg)
+
+    default_image_path = 'image/DJI_0142_frame_0041.jpg'
+    st.header('Please upload an image')
+    file = st.file_uploader('', type=['png', 'jpg', 'jpeg'])
+
+    if file:
+        image = Image.open(file)
+        st.image(image, caption="Uploaded Image", use_column_width=True)
+
+        click_instance_segmentation(image, model_selection, predictor)        
+
+    else:
+        st.write("No image uploaded. Using default example image.")
+        image = Image.open(default_image_path)
+        st.image(image, caption="Default Example Image", use_column_width=True)
+
+        click_instance_segmentation(image, model_selection, predictor)

output/obj_train/cfg.pickle

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f51e1c968bc79c9529217e626bdf89f84ac0cb73810377f9c20af8506e32e566
+size 5314

output/obj_train/coco_instances_results.json

@@ -0,0 +1 @@
+[{"image_id": 0, "category_id": 1, "bbox": [1380.3045654296875, 0.0, 1438.6485595703125, 1592.285400390625], "score": 0.9805574417114258}, {"image_id": 1, "category_id": 1, "bbox": [1874.0032958984375, 978.786376953125, 978.5140380859375, 1117.531005859375], "score": 0.9970454573631287}, {"image_id": 2, "category_id": 1, "bbox": [3113.146484375, 759.0662841796875, 691.964599609375, 623.3697509765625], "score": 0.9902016520500183}, {"image_id": 2, "category_id": 1, "bbox": [57.53554916381836, 633.5829467773438, 884.0471801757812, 1299.072021484375], "score": 0.14743082225322723}, {"image_id": 2, "category_id": 1, "bbox": [2850.850830078125, 511.6058044433594, 949.111572265625, 992.2442626953125], "score": 0.10244540870189667}, {"image_id": 4, "category_id": 1, "bbox": [0.0, 778.7610473632812, 626.103515625, 656.4393920898438], "score": 0.9994876384735107}, {"image_id": 5, "category_id": 1, "bbox": [1771.83935546875, 247.21034240722656, 1810.787841796875, 1463.2742919921875], "score": 0.9947130084037781}, {"image_id": 8, "category_id": 1, "bbox": [1909.879150390625, 1464.859375, 679.975341796875, 595.724365234375], "score": 0.9981029033660889}, {"image_id": 14, "category_id": 1, "bbox": [541.721923828125, 797.4324951171875, 697.9720458984375, 637.4742431640625], "score": 0.9980442523956299}, {"image_id": 15, "category_id": 1, "bbox": [1857.3994140625, 403.26446533203125, 1328.87353515625, 1625.130859375], "score": 0.9933504462242126}, {"image_id": 15, "category_id": 1, "bbox": [134.29571533203125, 1116.14013671875, 1241.58935546875, 1036.642578125], "score": 0.1047867015004158}, {"image_id": 16, "category_id": 1, "bbox": [2842.546630859375, 689.4129638671875, 704.650390625, 623.471435546875], "score": 0.9981980919837952}, {"image_id": 18, "category_id": 1, "bbox": [1617.198974609375, 1239.3822021484375, 680.04638671875, 751.341796875], "score": 0.9980385899543762}, {"image_id": 20, "category_id": 1, "bbox": [1600.5550537109375, 757.17236328125, 540.1990966796875, 640.1346435546875], "score": 0.9994283318519592}, {"image_id": 23, "category_id": 1, "bbox": [1385.652587890625, 0.0, 1457.248779296875, 1388.0927734375], "score": 0.9772377610206604}, {"image_id": 23, "category_id": 1, "bbox": [171.53965759277344, 777.9434814453125, 1182.2618408203125, 1382.0565185546875], "score": 0.12595009803771973}]

output/obj_train/events.out.tfevents.1697765063.nl869hipei.66.0

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:85b024fbd24535d7aaa02f0b8afec0096e9ca0b6563527bb6b8cd9f0183dfa93
+size 230768

output/obj_train/last_checkpoint

@@ -0,0 +1 @@
+model_final.pth

output/obj_train/model_best.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c46f433319ae3bd5872fb31c8a05671b3df7d7be0a5b150e88d5a8aa339c1054
+size 330024803

output/pointrend/.gitattributes

@@ -0,0 +1,2 @@
+*.pth filter=lfs diff=lfs merge=lfs -text
+detectron2_repo filter=lfs diff=lfs merge=lfs -text

output/pointrend/cfg.pickle

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4f2b3ef2d86630cade3f212f8fc5b35044416e5bd310236bf0b5227e8b1c1781
+size 5815

output/pointrend/events.out.tfevents.1696544539.n5kgwzlg1r.444.0

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6edf5ccc8576db52b9c29d91b116bcfce0bb36d06950317fa8bb382f7a1daaab
+size 150

output/pointrend/events.out.tfevents.1696544615.n5kgwzlg1r.532.0

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8ecf31f5ee105fa94446830204f394748310a13a19e28601289caea758d3b8c8
+size 150

output/pointrend/events.out.tfevents.1696544818.n5kgwzlg1r.636.0

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e2c33e01706eebbaeac2c530798e52f6f1ed48d9c32c43bad453a4270575daad
+size 150

output/pointrend/events.out.tfevents.1696545047.n5kgwzlg1r.726.0

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e9c638b35931841b76b9eb49ba1eddd1c20a21cfca19d18f0e46fe3c6d48c31d
+size 150

output/pointrend/events.out.tfevents.1696545254.n5kgwzlg1r.792.0

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:276913e82deffbff4d6d75bc7d3484e7659d7e5d86560090606aac7ffab776e5
+size 320382

output/pointrend/last_checkpoint

@@ -0,0 +1 @@
+model_final.pth

output/pointrend/model_best.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b33e1d8fb2347dab4b172dff25821bea64aa88b501a6b457f5eb7b85ea571e13
+size 445740323

output/pointrend_4_cls_7k/cfg.pickle

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:52736bfe061572f53f16bba1bd129936a5dad47346e28667db20cfd63e9dcc61
+size 5824

output/pointrend_4_cls_7k/events.out.tfevents.1697558440.n3vd74xx22.59.0

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:deec6381f3dc2aab1f6cf1426a34c074a866b4a5542dfe93494f8fb5d3842596
+size 322462

output/pointrend_4_cls_7k/events.out.tfevents.1697600075.n8pym182e6.59.0

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0c038a073127625381eac610e38c0aa81d7d24f5ec1b54357d7c9d9733db6fce
+size 451486

output/pointrend_4_cls_7k/last_checkpoint

@@ -0,0 +1 @@
+model_final.pth

output/pointrend_4_cls_7k/model_best.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b458e0ee5e84438d79ab781e9474412ee26a864daaf94778538210dfcef24b0e
+size 446642467

requirements.txt

@@ -0,0 +1,12 @@
+numpy
+pandas
+Pillow
+opencv-python
+opencv-python-headless
+torch
+torchvision
+streamlit
+streamlit-option-menu
+streamlit-aggrid
+
+

viz_app.py

@@ -0,0 +1,26 @@
+import streamlit as st
+from detectron2.data import Metadata, MetadataCatalog, DatasetCatalog
+from detectron2.utils.visualizer import Visualizer
+
+
+# Define the plot_single_image function
+def plot_single_image(im, predictor, num_classes):
+    
+    outputs = predictor(im)
+
+    if num_classes == 2:
+        modified_metadata = Metadata()
+        modified_metadata.thing_classes = ["Nest", "Inorganic_material"]
+    elif num_classes == 4:
+        modified_metadata = Metadata()
+        modified_metadata.thing_classes = ["Plastic", "Fishing_Net", "Rope/Cloth", "Nest"]
+        
+    st.write('Using Visualizer to draw the predictions on Image')
+    v = Visualizer(im[:, :, ::-1], modified_metadata, scale=1.2)
+    out = v.draw_instance_predictions(outputs["instances"].to("cpu"))
+    st.image(out.get_image()[:, :, ::-1])
+
+    return outputs
+
+
+