add make dataset

2026-02-08 20:21:57 +09:00
parent 5bfde5798f
commit 045e3da923
101 changed files with 4039 additions and 0 deletions
--- a/kamco-make-dataset-generation/src/main/resources/application.yml
+++ b/kamco-make-dataset-generation/src/main/resources/application.yml
@@ -0,0 +1,33 @@
+spring:
+  application:
+    name: kamco-geojson-scheduler
+  profiles:
+    active: local
+  datasource:
+    driver-class-name: org.postgresql.Driver
+    hikari:
+      minimum-idle: 2
+      maximum-pool-size: 2
+      connection-timeout: 20000
+      idle-timeout: 300000
+      max-lifetime: 1800000
+  batch:
+    job:
+      enabled: true
+    initialize-schema: never
+
+training-data:
+  docker:
+    image: kamco-cd-dataset:latest
+    user: "1000:1000"
+    dataset-volume: /kamco-nfs/dataset:/dataset
+    images-volume: /kamco-nfs/images:/kamco-nfs:ro
+    input-root: /dataset
+    output-root: /dataset
+    patch-size: 512
+    overlap-pct: 50
+    train-val-test-ratio:
+      - "0.7"
+      - "0.2"
+      - "0.1"
+    keep-empty-ratio: 0.1