initial commit

.gitignore

@@ -0,0 +1,18 @@
+.DS_Store
+flagged
+images
+labels
+labels-prev
+project*
+test*
+val
+yolo-labels
+ahmad.pt
+annotations.csv
+convert.py
+plot.py
+README.md
+LICENSE
+*.zip
+tf*
+weights/*

app.py

@@ -0,0 +1,65 @@
+from typing import Tuple
+from ultralytics import YOLO
+from ultralytics.engine.results import Boxes
+from ultralytics.utils.plotting import Annotator
+
+import gradio as gr
+
+cell_detector = YOLO("./weights/yolo_uninfected_cells.pt")
+redetr_detector = YOLO("./weights/redetr_infected_cells.pt")
+yolo_detector = YOLO("./weights/yolo_infected_cells.pt")
+
+models = {"Yolo V11": yolo_detector, "Real Time Detection Transformer": redetr_detector}
+classes = {"Yolo V11": [0], "Real Time Detection Transformer": [1]}
+
+
+def inference(image, model) -> Tuple[str, str, str]:
+    bboxes = []
+    labels = []
+    healthy_cell_count = 0
+    unhealthy_cell_count = 0
+    cells_results = cell_detector.predict(image, conf=0.5)
+    selected_model_results = models[model].predict(
+        image, conf=0.05, classes=classes[model], imgsz=1024
+    )
+
+    for cell_result in cells_results:
+        boxes: Boxes = cell_result.boxes
+        healthy_cells_bboxes = boxes.xyxy.tolist()
+        healthy_cell_count += len(healthy_cells_bboxes)
+        bboxes.extend(healthy_cells_bboxes)
+        labels.extend(["healthy"] * healthy_cell_count)
+
+    for res in selected_model_results:
+        boxes: Boxes = res.boxes
+        unhealthy_cells_bboxes = boxes.xyxy.tolist()
+        unhealthy_cell_count += len(unhealthy_cells_bboxes)
+        bboxes.extend(unhealthy_cells_bboxes)
+        labels.extend(["unhealthy"] * unhealthy_cell_count)
+
+    annotator = Annotator(image, font_size=5, line_width=1)
+
+    for box, label in zip(bboxes, labels):
+        annotator.box_label(box, label)
+
+    img = annotator.result()
+    return (img, healthy_cell_count, unhealthy_cell_count)
+
+
+ifer = gr.Interface(
+    fn=inference,
+    inputs=[
+        gr.Image(label="Input Image", type="numpy"),
+        gr.Dropdown(
+            choices=["Yolo V11", "Real Time Detection Transformer"], multiselect=False
+        ),
+    ],
+    outputs=[
+        gr.Image(label="Output Image", type="numpy"),
+        gr.Textbox(label="Healthy Cells Count"),
+        gr.Textbox(label="Infected Cells Count"),
+    ],
+    title="Blood Cancer Cell Detection and Counting"
+)
+
+ifer.launch(share=True)

data.yaml

@@ -0,0 +1,6 @@
+train: ./dataset/autosplit_train.txt
+val: ./dataset/autosplit_val.txt
+test: ./dataset/autosplit_test.txt
+nc: 1
+names:
+- cell

split.py

@@ -0,0 +1,7 @@
+from ultralytics.data.utils import autosplit
+
+autosplit(  
+    path="./dataset/images",
+    weights=(0.8, 0.1, 0.1),  # (train, validation, test) fractional splits
+    annotated_only=False,  # split only images with annotation file when True
+)

train.py

@@ -0,0 +1,6 @@
+from ultralytics import YOLO
+
+
+model = YOLO("yolov8n.pt")
+model.to('cuda')
+model.train(data="data.yaml", epochs=100, batch=8, device="cuda", imgsz=640)

val.py

@@ -0,0 +1,6 @@
+from ultralytics import YOLO
+
+
+model = YOLO("best_BCCM.pt")
+model.to('cuda')
+model.val(data="data.yaml")