.

Author: RootShell-coder

.github/workflows/docker.yml

@@ -0,0 +1,67 @@
+name: Build and check image
+on:
+  push:
+    branches: ["master"]
+    paths:
+      - "docker/**"
+      - ".github/workflows/**"
+  pull_request:
+    branches: ["master"]
+    paths:
+      - "docker/**"
+      - ".github/workflows/**"
+
+env:
+  REGISTRY: ghcr.io
+  LATEST_TAG: ghcr.io/rootshell-coder/person-detector:latest
+  TRIVY_SECURITY: SECURITY.txt
+
+jobs:
+  build:
+    name: Publish to github registry
+    runs-on: [ubuntu-24.04]
+    steps:
+      - name: Checkout
+        uses: actions/checkout@v4
+      - name: Set up QEMU
+        uses: docker/setup-qemu-action@v3
+      - name: Set up Docker Buildx
+        uses: docker/setup-buildx-action@v3
+      - name: Login to GHCR
+        uses: docker/login-action@v3
+        with:
+          registry: ${{ env.REGISTRY }}
+          username: ${{ github.actor }}
+          password: ${{ secrets.GITHUB_TOKEN }}
+      - name: Build and push
+        uses: docker/build-push-action@v5
+        with:
+          context: docker
+          push: true
+          platforms: linux/amd64,linux/arm64
+          tags: |
+            ${{ env.LATEST_TAG }}
+
+  check:
+    needs: [build]
+    name: Pull image
+    runs-on: [ubuntu-24.04]
+    steps:
+      - uses: actions/checkout@v4
+      - name: Create report file
+        run: |
+          echo "BUILD TIME" > ${{ env.VER_LOG }}
+          date +%s >> ${{ env.VER_LOG }}
+          echo "---" >> ${{ env.VER_LOG }}
+      - name: Check
+        run: |
+          docker pull ${{ env.LATEST_TAG }}
+          docker run --rm ghcr.io/rootshell-coder/trivy-cached:latest image ${{ env.LATEST_TAG }} --skip-db-update > ${{ env.TRIVY_SECURITY }}
+
+      - name: Commit changes
+        uses: EndBug/add-and-commit@v9
+        with:
+          author_name: RootShell-coder
+          author_email: Root.Shelling@gmail.com
+          message: "trivy"
+          add: "*.txt"

README.md

@@ -1 +1,44 @@
-# Person-detector
+# Faster R-CNN ResNet-50 с FPN (Feature Pyramid Network)
+
+ResNet-50 is a convolutional neural network (CNN) known for its deep architecture and efficiency. It excels at extracting features from images, which is key for object detection tasks.
+
+When you call `fasterrcnn_resnet50_fpn` in `torchvision`, the model automatically loads the pre-trained weights (if `pretrained=True` is specified), allowing you to use it "out of the box" for object detection tasks in images
+
+docker compose
+
+```yml
+---
+services:
+  coco2:
+    image: "person-detector"
+    runtime: "nvidia-runc"
+    ports:
+      - "5000:5000"
+    networks:
+      - "pd-net"
+    restart: "always"
+    deploy:
+      resources:
+        limits:
+          cpus: "6.0"
+          memory: "2G"
+        reservations:
+          cpus: "1.0"
+          memory: "2G"
+networks:
+  pd-net:
+    name: "pd-net"
+```
+
+Check my wedding photo directory and select all where a person is present.
+
+```bash
+photo_dir=/home/RootShell-coder/wedding
+
+for file in $(find ${photo_dir} -type f -name "*.jpg"); do
+  echo "Sending $file"
+  curl -s -X POST -F "file=@$file" -F "full_path=$(pwd)/$file" http://localhost:5000/detect
+done
+```
+
+In response, we get a JSON that can already be partically and sort photos

docker/.dockerignore

@@ -0,0 +1,2 @@
+.dockerignore
+Dockerfile

docker/Dockerfile

@@ -0,0 +1,11 @@
+FROM python:3.12.8-slim-bookworm
+
+WORKDIR /app
+
+COPY . .
+RUN set -eux; \
+  pip install torch torchvision matplotlib flask numpy Pillow
+
+EXPOSE 5000
+ENTRYPOINT []
+CMD ["python3", "app.py"]

docker/app.py

@@ -0,0 +1,101 @@
+import torch
+from torchvision import models, transforms
+from flask import Flask, request, jsonify
+from PIL import Image
+import numpy as np
+from io import BytesIO
+
+model = models.detection.fasterrcnn_resnet50_fpn(pretrained=True)
+model.eval()
+
+transform = transforms.Compose([
+    transforms.ToTensor()
+])
+
+def load_image_from_bytes(image_bytes):
+    try:
+        img = Image.open(BytesIO(image_bytes))
+
+        if img.format != 'JPEG':
+            return None, "Only JPEG images are allowed"
+
+        img = np.array(img)
+        return img, None
+    except Exception as e:
+        return None, str(e)
+
+def detect_person(image_np):
+    try:
+        image_tensor = transform(image_np).unsqueeze(0)
+
+        device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
+        model.to(device)
+        image_tensor = image_tensor.to(device)
+
+        with torch.no_grad():
+            prediction = model(image_tensor)
+
+        boxes = prediction[0]['boxes'].cpu().numpy()
+        labels = prediction[0]['labels'].cpu().numpy()
+        scores = prediction[0]['scores'].cpu().numpy()
+
+        detected_people = []
+
+        for i in range(len(scores)):
+            if scores[i] > 0.6 and labels[i] == 1:
+                confidence = float(scores[i]) * 100
+                detected_people.append({
+                    'confidence': round(confidence, 2),
+                    'box': boxes[i].tolist()
+                })
+
+        return detected_people, None
+    except Exception as e:
+        return None, str(e)
+
+app = Flask(__name__)
+
+@app.route('/detect', methods=['POST'])
+def detect():
+    try:
+        if 'file' not in request.files:
+            return jsonify({"error": "No file part"}), 400
+
+        file = request.files['file']
+
+        if file.filename == '':
+            return jsonify({"error": "No selected file"}), 400
+
+        # Проверяем MIME-тип файла
+        if file.content_type != 'image/jpeg':
+            return jsonify({"error": "Only JPEG images are allowed"}), 400
+
+        image_bytes = file.read()
+
+        image_np, error = load_image_from_bytes(image_bytes)
+        if image_np is None:
+            return jsonify({"error": error}), 400
+
+        detected_people, error = detect_person(image_np)
+        if detected_people is None:
+            return jsonify({"error": f"Error during detection: {error}"}), 500
+
+        full_path = request.form.get('full_path', None)
+
+        if detected_people:
+            return jsonify({
+                "message": "Person(s) detected on the image",
+                "people": detected_people,
+                "full_path": full_path
+            }), 200
+        else:
+            return jsonify({
+                "message": "No person detected on the image",
+                "full_path": full_path
+            }), 200
+
+    except Exception as e:
+        return jsonify({"error": f"An unexpected error occurred: {str(e)}"}), 500
+
+if __name__ == '__main__':
+    app.run(debug=True, host='0.0.0.0', port=5000, threaded=True)