Merge pull request #58 from FR-DC/0.0.11

0.0.11

.gitattributes

@@ -1 +1,3 @@
-Dockerfile text=auto eol=lf
+Dockerfile text=auto eol=lf
+* text=auto
+*.ipynb -linguist-detectable

.github/workflows/basic-tests.yml

@@ -4,6 +4,8 @@ on:
   push:
     branches: [ "main" ]
   pull_request:
+    paths:
+      - "src/frdc/**"
 
 jobs:
   build:

.github/workflows/build-ls-image.yml

@@ -0,0 +1,47 @@
+name: Publish Custom Label Studio Image
+on:
+  push:
+    branches: [ 'terraform' ]
+    paths:
+      - 'src/terraform/Dockerfile'
+  workflow_dispatch:
+
+env:
+  REGISTRY: ghcr.io
+  IMAGE_NAME: ${{ github.repository }}
+
+jobs:
+  build-and-push-image:
+    runs-on: ubuntu-latest
+    # Sets the permissions granted to the `GITHUB_TOKEN` for the actions in this job.
+    permissions:
+      contents: read
+      packages: write
+
+    steps:
+      - name: Checkout repository
+        uses: actions/checkout@v4
+
+      - name: Log in to the Container registry
+        uses: docker/[email protected]
+        with:
+          registry: ${{ env.REGISTRY }}
+          username: ${{ github.actor }}
+          password: ${{ secrets.GITHUB_TOKEN }}
+
+      - name: Extract metadata (tags, labels) for Docker
+        id: meta
+        uses: docker/[email protected]
+        with:
+          images: ${{ env.REGISTRY }}/${{ env.IMAGE_NAME }}
+
+      # This step uses the `docker/build-push-action` action to build the image, based on your repository's `Dockerfile`. If the build succeeds, it pushes the image to GitHub Packages.
+      # It uses the `context` parameter to define the build's context as the set of files located in the specified path. For more information, see "[Usage](https://github.com/docker/build-push-action#usage)" in the README of the `docker/build-push-action` repository.
+      # It uses the `tags` and `labels` parameters to tag and label the image with the output from the "meta" step.
+      - name: Build and push Docker image
+        uses: docker/[email protected]
+        with:
+          context: ./src/terraform
+          push: true
+          tags: ${{ steps.meta.outputs.tags }}
+          labels: ${{ steps.meta.outputs.labels }}

.gitignore

@@ -161,6 +161,43 @@ cython_debug/
 # Reference: https://intellij-support.jetbrains.com/hc/en-us/articles/206544839
 .idea/
 
+# Terraform Ignores
+# Local .terraform directories
+**/.terraform/*
+
+# .tfstate files
+*.tfstate
+*.tfstate.*
+
+# Crash log files
+crash.log
+crash.*.log
+
+# Exclude all .tfvars files, which are likely to contain sensitive data, such as
+# password, private keys, and other secrets. These should not be part of version
+# control as they are data points which are potentially sensitive and subject
+# to change depending on the environment.
+*.tfvars
+*.tfvars.json
+
+# Ignore override files as they are usually used to override resources locally and so
+# are not checked in
+override.tf
+override.tf.json
+*_override.tf
+*_override.tf.json
+
+# Include override files you do wish to add to version control using negated pattern
+# !example_override.tf
+
+# Include tfplan files to ignore the plan output of command: terraform plan -out=tfplan
+# example: *tfplan*
+
+# Ignore CLI configuration files
+.terraformrc
+terraform.rc
+
+
 # Ignores the raw .tif files
 rsc/**/*.tif
 

notebooks/README.md

@@ -0,0 +1,19 @@
+# Jupyter Notebooks
+
+## FRRD-60: Consistency as a feature discriminator for Novelty Detection
+
+Hypothesis: Consistency can be used as a measure to discriminate between normal
+and novel data. In the case of FRDC, it's a metric to separate seen and unseen
+tree-species. Seen data will have high consistency, while unseen data will have
+low consistency.
+
+Conclusion: It's not possible. Consistency is a measure of the similarity
+of output distributions, upon different pertubations of the input. A simple
+counter-example is the following: a black square in a white background.
+The consistency of that image is always perfect given weak augmentations 
+(flips). We also show that using CIFAR10 and noise datasets that the
+Jenson-Shannon divergence goes against our hypothesis, sometimes yielding
+higher consistency despite being Out-of-Distribution (OOD).
+
+Author Discussion: We believe that the formulated hypothesis, while on the 
+surface, plausible, requires more mathematical rigor to be proven.

notebooks/frrd-60/.gitignore

@@ -0,0 +1,2 @@
+data/
+*.ckpt

src/frdc/load/label_studio.py

@@ -16,35 +16,42 @@ class Task(dict):
     def get_bounds_and_labels(self) -> tuple[list[tuple[int, int]], list[str]]:
         bounds = []
         labels = []
-        for ann_ix, ann in enumerate(self["annotations"]):
-            results = ann["result"]
-            for r_ix, r in enumerate(results):
-                r: dict
-
-                # We flatten the value dict into the result dict
-                v = r.pop("value")
-                r = {**r, **v}
-
-                # Points are in percentage, we need to convert them to pixels
-                r["points"] = [
-                    (
-                        int(x * r["original_width"] / 100),
-                        int(y * r["original_height"] / 100),
-                    )
-                    for x, y in r["points"]
-                ]
-
-                # Only take the first label as this is not a multi-label task
-                r["label"] = r.pop("polygonlabels")[0]
-                if not r["closed"]:
-                    logger.warning(
-                        f"Label for {r['label']} @ {r['points']} not closed. "
-                        f"Skipping"
-                    )
-                    continue
-
-                bounds.append(r["points"])
-                labels.append(r["label"])
+
+        # for ann_ix, ann in enumerate(self["annotations"]):
+
+        ann = self["annotations"][0]
+        results = ann["result"]
+        for r_ix, r in enumerate(results):
+            r: dict
+
+            # See Issue FRML-78: Somehow some labels are actually just metadata
+            if r["from_name"] != "label":
+                continue
+
+            # We flatten the value dict into the result dict
+            v = r.pop("value")
+            r = {**r, **v}
+
+            # Points are in percentage, we need to convert them to pixels
+            r["points"] = [
+                (
+                    int(x * r["original_width"] / 100),
+                    int(y * r["original_height"] / 100),
+                )
+                for x, y in r["points"]
+            ]
+
+            # Only take the first label as this is not a multi-label task
+            r["label"] = r.pop("polygonlabels")[0]
+            if not r["closed"]:
+                logger.warning(
+                    f"Label for {r['label']} @ {r['points']} not closed. "
+                    f"Skipping"
+                )
+                continue
+
+            bounds.append(r["points"])
+            labels.append(r["label"])
 
         return bounds, labels
 

src/terraform/Dockerfile

@@ -0,0 +1,17 @@
+FROM ubuntu:22.04
+WORKDIR /app
+
+RUN apt-get update && apt-get install -y python3-pip
+
+RUN pip3 install label-studio
+
+# NOTE
+# This doesn't automatically port forward the port
+# to the host machine. You need to do that manually
+EXPOSE 8080
+
+ENV DJANGO_DB=default
+ENV POSTGRE_NAME=postgres
+ENV POSTGRE_PORT=5432
+
+ENTRYPOINT ["label-studio"]

src/terraform/google_compute_engine.tf

@@ -0,0 +1,101 @@
+# This code is compatible with Terraform 4.25.0 and versions that are backward compatible to 4.25.0.
+# For information about validating this Terraform code, see https://developer.hashicorp.com/terraform/tutorials/gcp-get-started/google-cloud-platform-build#format-and-validate-the-configuration
+
+resource "google_compute_instance" "label-studio" {
+  boot_disk {
+    auto_delete = true
+    device_name = var.google_compute_name
+
+    initialize_params {
+      image = "projects/cos-cloud/global/images/cos-stable-109-17800-147-38"
+      size  = 10
+      type  = "pd-balanced"
+    }
+
+    mode = "READ_WRITE"
+  }
+
+  can_ip_forward      = false
+  deletion_protection = false
+  enable_display      = false
+
+  labels = {
+    container-vm = "cos-stable-109-17800-147-38"
+    goog-ec-src  = "vm_add-tf"
+  }
+
+  machine_type = var.google_compute_machine_type
+
+  metadata = {
+    enable-oslogin            = "true"
+    gce-container-declaration = <<-EOF
+spec:
+    containers:
+    - name: label-studio
+      image: ghcr.io/fr-dc/frdc-ml:terraform
+      env:
+      - name: POSTGRE_PASSWORD
+        value: ${var.db_password}
+      - name: POSTGRE_USER
+        value: postgres.${supabase_project.production.id}
+      - name: POSTGRE_HOST
+        value: aws-0-${supabase_project.production.region}.pooler.supabase.com
+      - name: LABEL_STUDIO_DISABLE_SIGNUP_WITHOUT_LINK
+        value: true
+      - name: LABEL_STUDIO_USERNAME
+        value: ${var.ls_username}
+      - name: LABEL_STUDIO_PASSWORD
+        value: ${var.ls_password}
+      stdin: false
+      tty: false
+    restartPolicy: Always
+EOF
+  }
+
+  name = var.google_compute_name
+  zone = var.google_zone
+
+  network_interface {
+    access_config {
+      nat_ip       = google_compute_address.label-studio-ip.address
+      network_tier = "STANDARD"
+    }
+
+    queue_count = 0
+    stack_type  = "IPV4_ONLY"
+    subnetwork  = google_compute_network.label-studio-vpc.name
+  }
+
+  scheduling {
+    automatic_restart   = false
+    on_host_maintenance = "TERMINATE"
+    preemptible         = true
+    provisioning_model  = "SPOT"
+  }
+
+  service_account {
+    email = "[email protected]"
+    scopes = [
+      "https://www.googleapis.com/auth/devstorage.read_only",
+      "https://www.googleapis.com/auth/logging.write",
+      "https://www.googleapis.com/auth/monitoring.write",
+      "https://www.googleapis.com/auth/service.management.readonly",
+      "https://www.googleapis.com/auth/servicecontrol",
+      "https://www.googleapis.com/auth/trace.append"
+    ]
+  }
+
+  shielded_instance_config {
+    enable_integrity_monitoring = true
+    enable_secure_boot          = false
+    enable_vtpm                 = true
+  }
+
+  tags = ["label-studio"]
+  depends_on = [
+    google_compute_network.label-studio-vpc,
+    google_compute_address.label-studio-ip,
+    supabase_project.production
+  ]
+}
+

src/terraform/google_network.tf

@@ -0,0 +1,24 @@
+resource "google_compute_firewall" "label-studio-port" {
+  name    = var.google_firewall_name
+  network = google_compute_network.label-studio-vpc.name
+
+  allow {
+    protocol = "tcp"
+    ports    = ["22", "80", "443", "8080"]
+  }
+
+  source_ranges = ["0.0.0.0/0"]
+  target_tags   = ["label-studio"]
+  depends_on = [
+    google_compute_network.label-studio-vpc
+  ]
+}
+
+resource "google_compute_network" "label-studio-vpc" {
+  name                    = var.google_vpc_name
+  auto_create_subnetworks = true
+}
+
+resource "google_compute_address" "label-studio-ip" {
+  name = var.google_address_name
+}