CV / YOLOv8

YOLOv8m_seg

Version standard de la série de segmentation d'instance YOLOv8m_Seg, légère et efficace, conçue pour la segmentation au niveau pixel et la prédiction de masques avec un partitionnement régional précis, idéale pour le déploiement sur l'edge.

264 téléchargements

Taille

~35MB

Mémoire

1GB+

Précision

INT8

Choisissez l'appareil que vous utilisez. Le guide de configuration et la documentation seront mis à jour en conséquence.

Plateforme

Moteur d'inférence

Pour commencer

Déployer

sudo docker run --rm \
  --name rk3588-yolov8m-seg \
  --privileged \
  --net=host \
  -e PYTHONUNBUFFERED=1 \
  -e RKNN_LOG_LEVEL=0 \
  --device /dev/video1:/dev/video1 \
  --device /dev/dri/renderD129:/dev/dri/renderD129 \
  -v /proc/device-tree/compatible:/proc/device-tree/compatible \
  ghcr.io/seeed-projects/recomputer-rk-cv/rk3588-yolov8_seg:latest \
  python web_detection.py --model_path model/yolov8m_seg.rknn --video video/test.mp4

API REST

Utilisez l'API REST pour exécuter l'inférence. Copiez les commandes ci-dessous.

Curl

curl -X POST "http://127.0.0.1:8000/api/models/yolov8_seg/predict" -F "realtime=true"
# Or without file parameters
curl -X POST "http://127.0.0.1:8000/api/models/yolov8_seg/predict"

Python

import requests
import json
resp = requests.post(
    "http://127.0.0.1:8000/api/models/yolov8_seg/predict",
    json={},
    timeout=30
)
result = resp.json()
print(json.dumps(result, indent=2, ensure_ascii=False))

Détails du modèle

Demarrage rapide

1. Installer Docker

Executez les commandes suivantes sur la carte de developpement pour installer Docker :

bash

# Download installation script
curl -fsSL https://get.docker.com -o get-docker.sh
# Install using Aliyun mirror source
sudo sh get-docker.sh --mirror Aliyun
# Start Docker and enable auto-start on boot
sudo systemctl enable docker
sudo systemctl start docker

2. Executer le projet (Une commande, apercu double mode)

Ce projet prend en charge l'apercu simultane via Interface graphique locale et Navigateur Web. Le programme detecte automatiquement l'environnement d'affichage et passe en mode Web si aucun ecran n'est connecte.

Etape A : Configurer les permissions d'affichage (Optionnel)

Si vous avez un moniteur connecte et souhaitez voir la fenetre localement :

bash

xhost +local:docker

Etape B : Execution en un clic

Pour RK3588 :

bash

sudo docker run --rm --privileged --net=host \
    -e PYTHONUNBUFFERED=1 \
    -e RKNN_LOG_LEVEL=0 \
    --device /dev/video0:/dev/video0 \
    --device /dev/dri/renderD128:/dev/dri/renderD128 \
    -v /proc/device-tree/compatible:/proc/device-tree/compatible \
    ghcr.io/seeed-projects/recomputer-rk-cv/rk3588-yolov8_seg:latest \
    python3 web_detection.py --model_path model/yolov8n_seg.rknn --video video/test.mp4

Pour RK3576 :

bash

sudo docker run --rm --privileged --net=host \
    -e PYTHONUNBUFFERED=1 \
    -e RKNN_LOG_LEVEL=0 \
    --device /dev/video0:/dev/video0 \
    --device /dev/dri/renderD128:/dev/dri/renderD128 \
    -v /proc/device-tree/compatible:/proc/device-tree/compatible \
    ghcr.io/seeed-projects/recomputer-rk-cv/rk3576-yolov8_seg:latest \
    python3 web_detection.py --model_path model/yolov8n_seg.rknn --video video/test.mp4

Acces via : http://<Board_IP>:8000

Note : Si vous avez besoin de classes personnalisees, vous pouvez ajouter le montage -v $(pwd)/class_config.txt:/app/class_config.txt \ et le parametre --class_path. Le programme utilise par defaut les 80 classes COCO.

Exemple :

bash

sudo docker run --rm --privileged --net=host \
    -e PYTHONUNBUFFERED=1 \
    -e RKNN_LOG_LEVEL=0 \
    -v $(pwd)/class_config.txt:/app/class_config.txt \
    --device /dev/video0:/dev/video0 \
    --device /dev/dri/renderD128:/dev/dri/renderD128 \
    -v /proc/device-tree/compatible:/proc/device-tree/compatible \
    ghcr.io/seeed-projects/recomputer-rk-cv/rk3576-yolov8_seg:latest \
    python3 web_detection.py --model_path model/yolov8n-seg.rknn --video video/test.mp4 --class_path class_config.txt

Note : Si vous souhaitez tester avec une video locale au lieu d'une camera, utilisez le parametre --video_path :

bash

sudo docker run --rm --privileged --net=host \
    -e PYTHONUNBUFFERED=1 \
    -e RKNN_LOG_LEVEL=0 \
    -v $(pwd)/video:/app/video \
    --device /dev/dri/renderD128:/dev/dri/renderD128 \
    -v /proc/device-tree/compatible:/proc/device-tree/compatible \
    ghcr.io/seeed-projects/recomputer-rk-cv/rk3576-yolov8_seg:latest \
    python3 web_detection.py --model_path model/yolov8n-seg.rknn --video_path video/test.mp4

🔌 Documentation de l'API

Ce projet fournit des interfaces RESTful compatibles avec la norme API Ultralytics Cloud, prenant en charge la segmentation d'instance via le telechargement d'images, de videos ou l'appel direct de la camera.

1. Interface d'inference du modele (Predict)

Point de terminaison : POST /api/models/yolov8_seg/predict (ou /api/models/yolo11/predict selon le mapping exact du script)

Parametres de la requete (Multipart/Form-Data) :

file : (Optionnel) Fichier image a detecter.
video : (Optionnel) Fichier video MP4 a detecter.
timestamp : (Optionnel) Horodatage dans le fichier video (secondes), retourne les resultats de detection pour la trame a ce point. La valeur par defaut est 0.
realtime : (Optionnel) Booleen. Si true ou si aucun parametre file/video n'est fourni, retourne les resultats de detection pour la trame actuelle de la camera.
conf : (Optionnel) Seuil de confiance pour une requete unique, plage 0.0-1.0.
iou : (Optionnel) Seuil IOU NMS pour une requete unique, plage 0.0-1.0.

Exemples d'utilisation :

1. Detection d'image :

bash

curl -X POST "http://127.0.0.1:8000/api/models/yolov8_seg/predict" -F "file=@/home/cat/001.jpg"

2. Detection de trame video specifique :

bash

curl -X POST "http://127.0.0.1:8000/api/models/yolov8_seg/predict" -F "video=@/home/cat/test.mp4" -F "timestamp=5.5"

3. Obtenir la detection de la trame camera actuelle :

bash

curl -X POST "http://127.0.0.1:8000/api/models/yolov8_seg/predict" -F "realtime=true"

Format de reponse (JSON) :

json

{
  "success": true,
  "source": "realtime camera frame",
  "predictions": [
    {
      "class": "cup",
      "confidence": 0.9840346574783325,
      "box": { "x1": 100, "y1": 200, "x2": 300, "y2": 500 },
      "polygons": [
        [{"x":558.0,"y":336.0},{"x":558.0,"y":767.0},{"x":560.0,"y":767.0},{"x":561.0,"y":768.0},{"x":561.0,"y":773.0},{"x":563.0,"y":773.0},{"x":564.0,"y":774.0},{"x":564.0,"y":782.0},{"x":566.0,"y":782.0},{"x":567.0,"y":783.0},{"x":567.0,"y":788.0},{"x":569.0,"y":788.0},{"x":570.0,"y":789.0},{"x":570.0,"y":791.0},{"x":572.0,"y":791.0},{"x":573.0,"y":792.0},{"x":573.0,"y":794.0},{"x":575.0,"y":794.0},{"x":576.0,"y":795.0},{"x":576.0,"y":806.0},{"x":578.0,"y":806.0},{"x":579.0,"y":807.0},{"x":579.0,"y":815.0},{"x":581.0,"y":815.0},{"x":582.0,"y":816.0},{"x":582.0,"y":824.0},{"x":584.0,"y":824.0},{"x":585.0,"y":825.0},{"x":585.0,"y":839.0},{"x":587.0,"y":839.0},{"x":588.0,"y":840.0},{"x":588.0,"y":854.0},{"x":590.0,"y":854.0},{"x":591.0,"y":855.0},{"x":591.0,"y":866.0},{"x":593.0,"y":866.0},{"x":594.0,"y":867.0},{"x":594.0,"y":875.0},{"x":596.0,"y":875.0},{"x":597.0,"y":876.0},{"x":597.0,"y":884.0},{"x":599.0,"y":884.0},{"x":600.0,"y":885.0},{"x":600.0,"y":893.0},{"x":602.0,"y":893.0},{"x":603.0,"y":894.0},{"x":603.0,"y":899.0},{"x":605.0,"y":899.0},{"x":606.0,"y":900.0},{"x":606.0,"y":902.0},{"x":608.0,"y":902.0},{"x":609.0,"y":903.0},{"x":609.0,"y":905.0},{"x":611.0,"y":905.0},{"x":612.0,"y":906.0},{"x":612.0,"y":908.0},{"x":614.0,"y":908.0},{"x":615.0,"y":909.0},{"x":615.0,"y":911.0},{"x":620.0,"y":911.0},{"x":621.0,"y":912.0},{"x":621.0,"y":917.0},{"x":623.0,"y":917.0},{"x":624.0,"y":918.0},{"x":624.0,"y":920.0},{"x":626.0,"y":920.0},{"x":627.0,"y":921.0},{"x":627.0,"y":923.0},{"x":629.0,"y":923.0},{"x":630.0,"y":924.0},{"x":630.0,"y":926.0},{"x":632.0,"y":926.0},{"x":633.0,"y":927.0},{"x":633.0,"y":929.0},{"x":635.0,"y":929.0},{"x":636.0,"y":930.0},{"x":636.0,"y":932.0},{"x":641.0,"y":932.0},{"x":642.0,"y":933.0},{"x":642.0,"y":935.0},{"x":644.0,"y":935.0},{"x":645.0,"y":936.0},{"x":645.0,"y":938.0},{"x":650.0,"y":938.0},{"x":651.0,"y":939.0},{"x":651.0,"y":944.0},{"x":656.0,"y":944.0},{"x":657.0,"y":945.0},{"x":657.0,"y":947.0},{"x":662.0,"y":947.0},{"x":663.0,"y":948.0},{"x":663.0,"y":950.0},{"x":665.0,"y":950.0},{"x":666.0,"y":951.0},{"x":666.0,"y":953.0},{"x":671.0,"y":953.0},{"x":672.0,"y":954.0},{"x":672.0,"y":956.0},{"x":677.0,"y":956.0},{"x":678.0,"y":957.0},{"x":678.0,"y":959.0},{"x":686.0,"y":959.0},{"x":687.0,"y":960.0},{"x":687.0,"y":962.0},{"x":695.0,"y":962.0},{"x":696.0,"y":963.0},{"x":696.0,"y":965.0},{"x":701.0,"y":965.0},{"x":702.0,"y":966.0},{"x":702.0,"y":968.0},{"x":713.0,"y":968.0},{"x":714.0,"y":969.0},{"x":714.0,"y":971.0},{"x":869.0,"y":971.0},{"x":869.0,"y":969.0},{"x":870.0,"y":968.0},{"x":884.0,"y":968.0},{"x":884.0,"y":966.0},{"x":885.0,"y":965.0},{"x":893.0,"y":965.0},{"x":893.0,"y":963.0},{"x":894.0,"y":962.0},{"x":899.0,"y":962.0},{"x":899.0,"y":960.0},{"x":900.0,"y":959.0},{"x":911.0,"y":959.0},{"x":911.0,"y":957.0},{"x":912.0,"y":956.0},{"x":917.0,"y":956.0},{"x":917.0,"y":954.0},{"x":918.0,"y":953.0},{"x":923.0,"y":953.0},{"x":923.0,"y":951.0},{"x":924.0,"y":950.0},{"x":929.0,"y":950.0},{"x":929.0,"y":948.0},{"x":930.0,"y":947.0},{"x":935.0,"y":947.0},{"x":935.0,"y":945.0},{"x":936.0,"y":944.0},{"x":941.0,"y":944.0},{"x":941.0,"y":942.0},{"x":942.0,"y":941.0},{"x":947.0,"y":941.0},{"x":947.0,"y":939.0},{"x":948.0,"y":938.0},{"x":953.0,"y":938.0},{"x":953.0,"y":936.0},{"x":954.0,"y":935.0},{"x":962.0,"y":935.0},{"x":962.0,"y":933.0},{"x":963.0,"y":932.0},{"x":968.0,"y":932.0},{"x":968.0,"y":930.0},{"x":969.0,"y":929.0},{"x":971.0,"y":929.0},{"x":971.0,"y":927.0},{"x":972.0,"y":926.0},{"x":977.0,"y":926.0},{"x":977.0,"y":924.0},{"x":978.0,"y":923.0},{"x":983.0,"y":923.0},{"x":983.0,"y":921.0},{"x":984.0,"y":920.0},{"x":989.0,"y":920.0},{"x":989.0,"y":918.0},{"x":990.0,"y":917.0},{"x":992.0,"y":917.0},{"x":992.0,"y":915.0},{"x":993.0,"y":914.0},{"x":998.0,"y":914.0},{"x":998.0,"y":912.0},{"x":999.0,"y":911.0},{"x":1004.0,"y":911.0},{"x":1004.0,"y":909.0},{"x":1005.0,"y":908.0},{"x":1007.0,"y":908.0},{"x":1007.0,"y":906.0},{"x":1008.0,"y":905.0},{"x":1010.0,"y":905.0},{"x":1010.0,"y":903.0},{"x":1011.0,"y":902.0},{"x":1013.0,"y":902.0},{"x":1013.0,"y":900.0},{"x":1014.0,"y":899.0},{"x":1019.0,"y":899.0},{"x":1019.0,"y":897.0},{"x":1020.0,"y":896.0},{"x":1022.0,"y":896.0},{"x":1022.0,"y":894.0},{"x":1023.0,"y":893.0},{"x":1025.0,"y":893.0},{"x":1025.0,"y":891.0},{"x":1026.0,"y":890.0},{"x":1028.0,"y":890.0},{"x":1028.0,"y":888.0},{"x":1029.0,"y":887.0},{"x":1031.0,"y":887.0},{"x":1031.0,"y":885.0},{"x":1032.0,"y":884.0},{"x":1034.0,"y":884.0},{"x":1034.0,"y":882.0},{"x":1035.0,"y":881.0},{"x":1037.0,"y":881.0},{"x":1037.0,"y":879.0},{"x":1038.0,"y":878.0},{"x":1040.0,"y":878.0},{"x":1040.0,"y":876.0},{"x":1041.0,"y":875.0},{"x":1043.0,"y":875.0},{"x":1043.0,"y":873.0},{"x":1044.0,"y":872.0},{"x":1046.0,"y":872.0},{"x":1046.0,"y":867.0},{"x":1047.0,"y":866.0},{"x":1049.0,"y":866.0},{"x":1049.0,"y":864.0},{"x":1050.0,"y":863.0},{"x":1052.0,"y":863.0},{"x":1052.0,"y":861.0},{"x":1053.0,"y":860.0},{"x":1055.0,"y":860.0},{"x":1055.0,"y":858.0},{"x":1056.0,"y":857.0},{"x":1058.0,"y":857.0},{"x":1058.0,"y":852.0},{"x":1059.0,"y":851.0},{"x":1061.0,"y":851.0},{"x":1061.0,"y":849.0},{"x":1062.0,"y":848.0},{"x":1064.0,"y":848.0},{"x":1064.0,"y":846.0},{"x":1065.0,"y":845.0},{"x":1067.0,"y":845.0},{"x":1067.0,"y":840.0},{"x":1068.0,"y":839.0},{"x":1070.0,"y":839.0},{"x":1070.0,"y":837.0},{"x":1071.0,"y":836.0},{"x":1073.0,"y":836.0},{"x":1073.0,"y":834.0},{"x":1074.0,"y":833.0},{"x":1076.0,"y":833.0},{"x":1076.0,"y":831.0},{"x":1077.0,"y":830.0},{"x":1085.0,"y":830.0},{"x":1085.0,"y":828.0},{"x":1086.0,"y":827.0},{"x":1094.0,"y":827.0},{"x":1094.0,"y":825.0},{"x":1095.0,"y":824.0},{"x":1103.0,"y":824.0},{"x":1103.0,"y":822.0},{"x":1104.0,"y":821.0},{"x":1109.0,"y":821.0},{"x":1109.0,"y":819.0},{"x":1110.0,"y":818.0},{"x":1121.0,"y":818.0},{"x":1121.0,"y":816.0},{"x":1122.0,"y":815.0},{"x":1136.0,"y":815.0},{"x":1136.0,"y":813.0},{"x":1137.0,"y":812.0},{"x":1142.0,"y":812.0},{"x":1142.0,"y":810.0},{"x":1143.0,"y":809.0},{"x":1148.0,"y":809.0},{"x":1148.0,"y":807.0},{"x":1149.0,"y":806.0},{"x":1154.0,"y":806.0},{"x":1154.0,"y":804.0},{"x":1155.0,"y":803.0},{"x":1163.0,"y":803.0},{"x":1163.0,"y":801.0},{"x":1164.0,"y":800.0},{"x":1166.0,"y":800.0},{"x":1166.0,"y":798.0},{"x":1167.0,"y":797.0},{"x":1172.0,"y":797.0},{"x":1172.0,"y":795.0},{"x":1173.0,"y":794.0},{"x":1175.0,"y":794.0},{"x":1175.0,"y":792.0},{"x":1176.0,"y":791.0},{"x":1181.0,"y":791.0},{"x":1181.0,"y":789.0},{"x":1182.0,"y":788.0},{"x":1187.0,"y":788.0},{"x":1187.0,"y":786.0},{"x":1188.0,"y":785.0},{"x":1190.0,"y":785.0},{"x":1190.0,"y":783.0},{"x":1191.0,"y":782.0},{"x":1196.0,"y":782.0},{"x":1196.0,"y":780.0},{"x":1197.0,"y":779.0},{"x":1202.0,"y":779.0},{"x":1202.0,"y":777.0},{"x":1203.0,"y":776.0},{"x":1205.0,"y":776.0},{"x":1205.0,"y":774.0},{"x":1206.0,"y":773.0},{"x":1208.0,"y":773.0},{"x":1208.0,"y":771.0},{"x":1209.0,"y":770.0},{"x":1214.0,"y":770.0},{"x":1214.0,"y":768.0},{"x":1215.0,"y":767.0},{"x":1217.0,"y":767.0},{"x":1217.0,"y":765.0},{"x":1218.0,"y":764.0},{"x":1223.0,"y":764.0},{"x":1223.0,"y":762.0},{"x":1224.0,"y":761.0},{"x":1226.0,"y":761.0},{"x":1226.0,"y":759.0},{"x":1227.0,"y":758.0},{"x":1229.0,"y":758.0},{"x":1229.0,"y":756.0},{"x":1230.0,"y":755.0},{"x":1235.0,"y":755.0},{"x":1235.0,"y":753.0},{"x":1236.0,"y":752.0},{"x":1238.0,"y":752.0},{"x":1238.0,"y":750.0},{"x":1239.0,"y":749.0},{"x":1241.0,"y":749.0},{"x":1241.0,"y":747.0},{"x":1242.0,"y":746.0},{"x":1244.0,"y":746.0},{"x":1244.0,"y":744.0},{"x":1245.0,"y":743.0},{"x":1247.0,"y":743.0},{"x":1247.0,"y":741.0},{"x":1248.0,"y":740.0},{"x":1250.0,"y":740.0},{"x":1250.0,"y":738.0},{"x":1251.0,"y":737.0},{"x":1253.0,"y":737.0},{"x":1253.0,"y":732.0},{"x":1254.0,"y":731.0},{"x":1256.0,"y":731.0},{"x":1256.0,"y":726.0},{"x":1257.0,"y":725.0},{"x":1259.0,"y":725.0},{"x":1259.0,"y":720.0},{"x":1260.0,"y":719.0},{"x":1262.0,"y":719.0},{"x":1262.0,"y":714.0},{"x":1263.0,"y":713.0},{"x":1265.0,"y":713.0},{"x":1265.0,"y":711.0},{"x":1266.0,"y":710.0},{"x":1268.0,"y":710.0},{"x":1268.0,"y":705.0},{"x":1269.0,"y":704.0},{"x":1271.0,"y":704.0},{"x":1271.0,"y":696.0},{"x":1272.0,"y":695.0},{"x":1274.0,"y":695.0},{"x":1274.0,"y":690.0},{"x":1275.0,"y":689.0},{"x":1277.0,"y":689.0},{"x":1277.0,"y":687.0},{"x":1278.0,"y":686.0},{"x":1280.0,"y":686.0},{"x":1280.0,"y":681.0},{"x":1281.0,"y":680.0},{"x":1283.0,"y":680.0},{"x":1283.0,"y":675.0},{"x":1284.0,"y":674.0},{"x":1286.0,"y":674.0},{"x":1286.0,"y":663.0},{"x":1287.0,"y":662.0},{"x":1289.0,"y":662.0},{"x":1289.0,"y":657.0},{"x":1290.0,"y":656.0},{"x":1292.0,"y":656.0},{"x":1292.0,"y":579.0},{"x":1290.0,"y":579.0},{"x":1289.0,"y":578.0},{"x":1289.0,"y":567.0},{"x":1287.0,"y":567.0},{"x":1286.0,"y":566.0},{"x":1286.0,"y":555.0},{"x":1284.0,"y":555.0},{"x":1283.0,"y":554.0},{"x":1283.0,"y":549.0},{"x":1281.0,"y":549.0},{"x":1280.0,"y":548.0},{"x":1280.0,"y":543.0},{"x":1278.0,"y":543.0},{"x":1277.0,"y":542.0},{"x":1277.0,"y":537.0},{"x":1275.0,"y":537.0},{"x":1274.0,"y":536.0},{"x":1274.0,"y":531.0},{"x":1272.0,"y":531.0},{"x":1271.0,"y":530.0},{"x":1271.0,"y":525.0},{"x":1269.0,"y":525.0},{"x":1268.0,"y":524.0},{"x":1268.0,"y":519.0},{"x":1266.0,"y":519.0},{"x":1265.0,"y":518.0},{"x":1265.0,"y":516.0},{"x":1263.0,"y":516.0},{"x":1262.0,"y":515.0},{"x":1262.0,"y":513.0},{"x":1260.0,"y":513.0},{"x":1259.0,"y":512.0},{"x":1259.0,"y":507.0},{"x":1257.0,"y":507.0},{"x":1256.0,"y":506.0},{"x":1256.0,"y":501.0},{"x":1254.0,"y":501.0},{"x":1253.0,"y":500.0},{"x":1253.0,"y":498.0},{"x":1251.0,"y":498.0},{"x":1250.0,"y":497.0},{"x":1250.0,"y":495.0},{"x":1248.0,"y":495.0},{"x":1247.0,"y":494.0},{"x":1247.0,"y":489.0},{"x":1245.0,"y":489.0},{"x":1244.0,"y":488.0},{"x":1244.0,"y":486.0},{"x":1242.0,"y":486.0},{"x":1241.0,"y":485.0},{"x":1241.0,"y":483.0},{"x":1239.0,"y":483.0},{"x":1238.0,"y":482.0},{"x":1238.0,"y":480.0},{"x":1236.0,"y":480.0},{"x":1235.0,"y":479.0},{"x":1235.0,"y":477.0},{"x":1233.0,"y":477.0},{"x":1232.0,"y":476.0},{"x":1232.0,"y":474.0},{"x":1230.0,"y":474.0},{"x":1229.0,"y":473.0},{"x":1229.0,"y":471.0},{"x":1227.0,"y":471.0},{"x":1226.0,"y":470.0},{"x":1226.0,"y":468.0},{"x":1224.0,"y":468.0},{"x":1223.0,"y":467.0},{"x":1223.0,"y":465.0},{"x":1221.0,"y":465.0},{"x":1220.0,"y":464.0},{"x":1220.0,"y":462.0},{"x":1218.0,"y":462.0},{"x":1217.0,"y":461.0},{"x":1217.0,"y":459.0},{"x":1212.0,"y":459.0},{"x":1211.0,"y":458.0},{"x":1211.0,"y":456.0},{"x":1209.0,"y":456.0},{"x":1208.0,"y":455.0},{"x":1208.0,"y":453.0},{"x":1206.0,"y":453.0},{"x":1205.0,"y":452.0},{"x":1205.0,"y":450.0},{"x":1203.0,"y":450.0},{"x":1202.0,"y":449.0},{"x":1202.0,"y":447.0},{"x":1197.0,"y":447.0},{"x":1196.0,"y":446.0},{"x":1196.0,"y":444.0},{"x":1191.0,"y":444.0},{"x":1190.0,"y":443.0},{"x":1190.0,"y":441.0},{"x":1188.0,"y":441.0},{"x":1187.0,"y":440.0},{"x":1187.0,"y":438.0},{"x":1185.0,"y":438.0},{"x":1184.0,"y":437.0},{"x":1184.0,"y":435.0},{"x":1182.0,"y":435.0},{"x":1181.0,"y":434.0},{"x":1181.0,"y":432.0},{"x":1176.0,"y":432.0},{"x":1175.0,"y":431.0},{"x":1175.0,"y":429.0},{"x":1173.0,"y":429.0},{"x":1172.0,"y":428.0},{"x":1172.0,"y":426.0},{"x":1170.0,"y":426.0},{"x":1169.0,"y":425.0},{"x":1169.0,"y":423.0},{"x":1164.0,"y":423.0},{"x":1163.0,"y":422.0},{"x":1163.0,"y":420.0},{"x":1155.0,"y":420.0},{"x":1154.0,"y":419.0},{"x":1154.0,"y":417.0},{"x":1149.0,"y":417.0},{"x":1148.0,"y":416.0},{"x":1148.0,"y":414.0},{"x":1143.0,"y":414.0},{"x":1142.0,"y":413.0},{"x":1142.0,"y":411.0},{"x":1137.0,"y":411.0},{"x":1136.0,"y":410.0},{"x":1136.0,"y":408.0},{"x":1128.0,"y":408.0},{"x":1127.0,"y":407.0},{"x":1127.0,"y":405.0},{"x":1122.0,"y":405.0},{"x":1121.0,"y":404.0},{"x":1121.0,"y":402.0},{"x":1116.0,"y":402.0},{"x":1115.0,"y":401.0},{"x":1115.0,"y":399.0},{"x":1110.0,"y":399.0},{"x":1109.0,"y":398.0},{"x":1109.0,"y":396.0},{"x":1101.0,"y":396.0},{"x":1100.0,"y":395.0},{"x":1100.0,"y":393.0},{"x":1095.0,"y":393.0},{"x":1094.0,"y":392.0},{"x":1094.0,"y":390.0},{"x":1092.0,"y":390.0},{"x":1091.0,"y":389.0},{"x":1091.0,"y":387.0},{"x":1089.0,"y":387.0},{"x":1088.0,"y":386.0},{"x":1088.0,"y":384.0},{"x":1083.0,"y":384.0},{"x":1082.0,"y":383.0},{"x":1082.0,"y":381.0},{"x":1080.0,"y":381.0},{"x":1079.0,"y":380.0},{"x":1079.0,"y":375.0},{"x":1074.0,"y":375.0},{"x":1073.0,"y":374.0},{"x":1073.0,"y":372.0},{"x":1065.0,"y":372.0},{"x":1064.0,"y":371.0},{"x":1064.0,"y":369.0},{"x":1059.0,"y":369.0},{"x":1058.0,"y":368.0},{"x":1058.0,"y":366.0},{"x":1050.0,"y":366.0},{"x":1049.0,"y":365.0},{"x":1049.0,"y":363.0},{"x":1032.0,"y":363.0},{"x":1031.0,"y":362.0},{"x":1031.0,"y":360.0},{"x":1002.0,"y":360.0},{"x":1001.0,"y":359.0},{"x":1001.0,"y":357.0},{"x":990.0,"y":357.0},{"x":989.0,"y":356.0},{"x":989.0,"y":354.0},{"x":981.0,"y":354.0},{"x":980.0,"y":353.0},{"x":980.0,"y":351.0},{"x":978.0,"y":351.0},{"x":977.0,"y":350.0},{"x":977.0,"y":348.0},{"x":972.0,"y":348.0},{"x":971.0,"y":347.0},{"x":971.0,"y":345.0},{"x":966.0,"y":345.0},{"x":965.0,"y":344.0},{"x":965.0,"y":342.0},{"x":954.0,"y":342.0},{"x":953.0,"y":341.0},{"x":953.0,"y":339.0},{"x":930.0,"y":339.0},{"x":929.0,"y":338.0},{"x":929.0,"y":336.0}]
      ]
    }
  ],
  "image": { "width": 1280, "height": 720 }
}

2. Interface de configuration systeme (Config)

Utilisee pour ajuster dynamiquement les seuils pour les flux video en temps reel et l'inference par defaut.

Obtenir la configuration actuelle

Point de terminaison : GET /api/config
Reponse : {"obj_thresh": 0.25, "nms_thresh": 0.45}

Mettre a jour la configuration systeme

Point de terminaison : POST /api/config
Corps de la requete (JSON) : {"obj_thresh": 0.3, "nms_thresh": 0.5}
Reponse : {"status": "success"}

3. Interface de flux video en temps reel (Video Feed)

Obtenez un flux video MJPEG en temps reel avec les boites de detection et les masques de segmentation dessines, pouvant etre directement integre dans les balises HTML <img>.

Point de terminaison : GET /api/video_feed
Exemple d'utilisation : <img src="http://<Board_IP>:8000/api/video_feed">

🛠️ Guide du developpeur (Recommandations de production)

Description du code

web_detection.py :
- Prise en charge double mode : Integre FastAPI, prenant en charge a la fois le rendu local et la sortie de streaming MJPEG.
- Adaptatif a l'environnement : Detecte automatiquement la variable d'environnement DISPLAY, ignorant silencieusement l'initialisation GUI si elle n'est pas presente.
- Inference RKNN : Encapsule l'initialisation RKNN, le chargement du modele et la logique d'inference multi-coeur.
- Chargement dynamique : Prend en charge le chargement dynamique de la configuration des classes via --class_path.
- Post-traitement : Decodage haute performance des boites englobantes base sur Numpy, NMS et extraction de contours (cv2.findContours).

Modification des modeles

Placez le modele .rknn entraine et converti dans le repertoire model/.
Ajoutez l'argument --model_path a la commande d'execution pour pointer vers le nouveau modele.