Update code to use the new ProcessAnnotatedImage job of biigle/largo

This also fixes #148 because the new job generates feature vectors, too.
biigle · Jan 19, 2024 · 82e375f · 82e375f
1 parent 0c9aa78
commit 82e375f
Show file tree

Hide file tree

Showing 15 changed files with 456 additions and 126 deletions.
diff --git a/src/Http/Controllers/Api/AnnotationCandidateController.php b/src/Http/Controllers/Api/AnnotationCandidateController.php
@@ -3,11 +3,11 @@
 namespace Biigle\Modules\Maia\Http\Controllers\Api;
 
 use Biigle\Http\Controllers\Api\Controller;
-use Biigle\Modules\Largo\Jobs\GenerateImageAnnotationPatch;
 use Biigle\Modules\Maia\AnnotationCandidateFeatureVector;
 use Biigle\Modules\Maia\Http\Requests\SubmitAnnotationCandidates;
 use Biigle\Modules\Maia\Http\Requests\UpdateAnnotationCandidate;
 use Biigle\Modules\Maia\Jobs\ConvertAnnotationCandidates;
+use Biigle\Modules\Maia\Jobs\ProcessObjectDetectedImage;
 use Biigle\Modules\Maia\MaiaJob;
 use Illuminate\Http\Response;
 use Pgvector\Laravel\Distance;
@@ -150,17 +150,21 @@ public function submit(SubmitAnnotationCandidates $request)
      */
     public function update(UpdateAnnotationCandidate $request)
     {
+        $candidate = $request->candidate;
         if ($request->filled('points')) {
-            $request->candidate->points = $request->input('points');
-            $disk = config('maia.annotation_candidate_storage_disk');
-            GenerateImageAnnotationPatch::dispatch($request->candidate, $disk)
+            $candidate->points = $request->input('points');
+            ProcessObjectDetectedImage::dispatch($candidate->image,
+                    only: [$candidate->id],
+                    maiaJob: $candidate->job,
+                    targetDisk: config('maia.annotation_candidate_storage_disk')
+                )
                 ->onQueue(config('largo.generate_annotation_patch_queue'));
         }
 
         if ($request->has('label_id')) {
-            $request->candidate->label_id = $request->input('label_id');
+            $candidate->label_id = $request->input('label_id');
         }
 
-        $request->candidate->save();
+        $candidate->save();
     }
 }
diff --git a/src/Http/Controllers/Api/TrainingProposalController.php b/src/Http/Controllers/Api/TrainingProposalController.php
@@ -3,10 +3,10 @@
 namespace Biigle\Modules\Maia\Http\Controllers\Api;
 
 use Biigle\Http\Controllers\Api\Controller;
-use Biigle\Modules\Largo\Jobs\GenerateImageAnnotationPatch;
 use Biigle\Modules\Maia\Events\MaiaJobContinued;
 use Biigle\Modules\Maia\Http\Requests\ContinueMaiaJob;
 use Biigle\Modules\Maia\Http\Requests\UpdateTrainingProposal;
+use Biigle\Modules\Maia\Jobs\ProcessNoveltyDetectedImage;
 use Biigle\Modules\Maia\MaiaJob;
 use Biigle\Modules\Maia\MaiaJobState as State;
 use Biigle\Modules\Maia\TrainingProposalFeatureVector;
@@ -149,14 +149,19 @@ public function submit(ContinueMaiaJob $request)
      */
     public function update(UpdateTrainingProposal $request)
     {
+        $proposal = $request->proposal;
         if ($request->filled('points')) {
-            $request->proposal->points = $request->input('points');
-            $disk = config('maia.training_proposal_storage_disk');
-            GenerateImageAnnotationPatch::dispatch($request->proposal, $disk)
+            $proposal->points = $request->input('points');
+            ProcessNoveltyDetectedImage::dispatch($proposal->image,
+                    only: [$proposal->id],
+                    maiaJob: $proposal->job,
+                    targetDisk: config('maia.training_proposal_storage_disk')
+                )
                 ->onQueue(config('largo.generate_annotation_patch_queue'));
         }
 
-        $request->proposal->selected = $request->input('selected', $request->proposal->selected);
-        $request->proposal->save();
+        $proposal->selected = $request->input('selected', $proposal->selected);
+
+        $proposal->save();
     }
 }
diff --git a/src/Jobs/ConvertAnnotationCandidates.php b/src/Jobs/ConvertAnnotationCandidates.php
@@ -5,12 +5,13 @@
 use Biigle\ImageAnnotation;
 use Biigle\ImageAnnotationLabel;
 use Biigle\Jobs\Job;
-use Biigle\Modules\Largo\Jobs\GenerateImageAnnotationPatch;
+use Biigle\Modules\Largo\Jobs\ProcessAnnotatedImage;
 use Biigle\Modules\Maia\MaiaJob;
 use Biigle\User;
 use Carbon\Carbon;
 use DB;
 use Illuminate\Queue\SerializesModels;
+use Illuminate\Support\Collection;
 
 class ConvertAnnotationCandidates extends Job
 {
@@ -47,9 +48,9 @@ class ConvertAnnotationCandidates extends Job
     /**
      * Newly created annotations.
      *
-     * @var array
+     * @var Collection
      */
-    protected $newAnnotations = [];
+    protected $newAnnotations;
 
     /**
      * Create a new isntance.
@@ -62,6 +63,7 @@ public function __construct(MaiaJob $job, User $user)
         $this->queue = config('maia.convert_annotations_queue');
         $this->job = $job;
         $this->user = $user;
+        $this->newAnnotations = collect([]);
     }
 
     /**
@@ -79,10 +81,15 @@ public function handle()
                     ->chunkById(10000, [$this, 'processChunk']);
             });
 
-            foreach ($this->newAnnotations as $annotation) {
-                GenerateImageAnnotationPatch::dispatch($annotation)
-                    ->onQueue(config('largo.generate_annotation_patch_queue'));
-            }
+            $this->newAnnotations
+                ->groupBy('image_id')
+                ->each(function ($group) {
+                    $image = $group[0]->image;
+                    $ids = $group->pluck('id')->all();
+                    ProcessAnnotatedImage::dispatch($image, only: $ids)
+                        ->onQueue(config('largo.generate_annotation_patch_queue'));
+                });
+
         } finally {
             $this->job->convertingCandidates = false;
             $this->job->save();

diff --git a/src/Jobs/GenerateAnnotationCandidatePatches.php b/src/Jobs/GenerateAnnotationCandidatePatches.php
@@ -2,23 +2,46 @@
 
 namespace Biigle\Modules\Maia\Jobs;
 
-class GenerateAnnotationCandidatePatches extends GenerateAnnotationPatches
+use Biigle\Jobs\Job;
+use Biigle\Modules\Maia\MaiaJob;
+use Illuminate\Contracts\Queue\ShouldQueue;
+use Illuminate\Queue\SerializesModels;
+
+class GenerateAnnotationCandidatePatches extends Job implements ShouldQueue
 {
+    use SerializesModels;
+
     /**
-     * Get a query for the annotations that have been created by this job.
+     * Ignore this job if the MAIA job does not exist any more.
      *
-     * @return \Illuminate\Database\Eloquent\Relations\BelongsTo
+     * @var bool
      */
-    protected function getCreatedAnnotations()
-    {
-        return $this->job->annotationCandidates();
-    }
+    protected $deleteWhenMissingModels = true;
 
     /**
-     * Get the storage disk to store the annotation patches to.
+     * Create a new isntance.
      */
-    protected function getPatchStorageDisk()
+    public function __construct(public MaiaJob $maiaJob)
+    {
+        //
+    }
+
+    public function handle(): void
     {
-        return config('maia.annotation_candidate_storage_disk');
+        $this->maiaJob->volume->images()
+            ->whereExists(fn ($q) =>
+                $q->select(\DB::raw(1))
+                    ->from('maia_annotation_candidates')
+                    ->where('maia_annotation_candidates.job_id', $this->maiaJob->id)
+                    ->whereColumn('maia_annotation_candidates.image_id', 'images.id')
+            )
+            ->eachById(fn ($image) =>
+                ProcessObjectDetectedImage::dispatch($image, $this->maiaJob,
+                        // Feature vectors are generated in a separate job on the GPU.
+                        skipFeatureVectors: true,
+                        targetDisk: config('maia.annotation_candidate_storage_disk')
+                    )
+                    ->onQueue(config('largo.generate_annotation_patch_queue'))
+            );
     }
 }
diff --git a/src/Jobs/GenerateAnnotationPatches.php b/src/Jobs/GenerateAnnotationPatches.php
diff --git a/src/Jobs/GenerateTrainingProposalPatches.php b/src/Jobs/GenerateTrainingProposalPatches.php
@@ -2,23 +2,46 @@
 
 namespace Biigle\Modules\Maia\Jobs;
 
-class GenerateTrainingProposalPatches extends GenerateAnnotationPatches
+use Biigle\Jobs\Job;
+use Biigle\Modules\Maia\MaiaJob;
+use Illuminate\Contracts\Queue\ShouldQueue;
+use Illuminate\Queue\SerializesModels;
+
+class GenerateTrainingProposalPatches extends Job implements ShouldQueue
 {
+    use SerializesModels;
+
     /**
-     * Get a query for the annotations that have been created by this job.
+     * Ignore this job if the MAIA job does not exist any more.
      *
-     * @return \Illuminate\Database\Eloquent\Relations\BelongsTo
+     * @var bool
      */
-    protected function getCreatedAnnotations()
-    {
-        return $this->job->trainingProposals();
-    }
+    protected $deleteWhenMissingModels = true;
 
     /**
-     * Get the storage disk to store the annotation patches to.
+     * Create a new isntance.
      */
-    protected function getPatchStorageDisk()
+    public function __construct(public MaiaJob $maiaJob)
+    {
+        //
+    }
+
+    public function handle(): void
     {
-        return config('maia.training_proposal_storage_disk');
+        $this->maiaJob->volume->images()
+            ->whereExists(fn ($q) =>
+                $q->select(\DB::raw(1))
+                    ->from('maia_training_proposals')
+                    ->where('maia_training_proposals.job_id', $this->maiaJob->id)
+                    ->whereColumn('maia_training_proposals.image_id', 'images.id')
+            )
+            ->eachById(fn ($image) =>
+                ProcessNoveltyDetectedImage::dispatch($image, $this->maiaJob,
+                        // Feature vectors are generated in a separate job on the GPU.
+                        skipFeatureVectors: true,
+                        targetDisk: config('maia.training_proposal_storage_disk')
+                    )
+                    ->onQueue(config('largo.generate_annotation_patch_queue'))
+            );
     }
 }
diff --git a/src/Jobs/ProcessNoveltyDetectedImage.php b/src/Jobs/ProcessNoveltyDetectedImage.php
@@ -0,0 +1,55 @@
+<?php
+
+namespace Biigle\Modules\Maia\Jobs;
+
+use Biigle\Modules\Largo\Jobs\ProcessAnnotatedImage;
+use Biigle\Modules\Maia\MaiaJob;
+use Biigle\Modules\Maia\TrainingProposal;
+use Biigle\Modules\Maia\TrainingProposalFeatureVector;
+use Biigle\VolumeFile;
+use Illuminate\Database\Eloquent\Builder;
+use Illuminate\Database\Eloquent\Collection;
+
+class ProcessNoveltyDetectedImage extends ProcessAnnotatedImage
+{
+    public function __construct(
+        public VolumeFile $file,
+        public MaiaJob $maiaJob,
+        public array $only = [],
+        public bool $skipFeatureVectors = false,
+        public ?string $targetDisk = null
+    )
+    {
+        parent::__construct($file, $only,
+            skipFeatureVectors: $skipFeatureVectors,
+            targetDisk: $targetDisk
+        );
+    }
+
+    /**
+     * {@inheritdoc}
+     */
+    protected function getAnnotationQuery(VolumeFile $file): Builder
+    {
+        return TrainingProposal::where('image_id', $file->id)
+            ->where('job_id', $this->maiaJob->id);
+    }
+
+    /**
+     * Create the feature vectors based on the Python script output.
+     */
+    protected function updateOrCreateFeatureVectors(Collection $annotations, \Generator $output): void
+    {
+        $annotations = $annotations->keyBy('id');
+        foreach ($output as $row) {
+            $annotation = $annotations->get($row[0]);
+            TrainingProposalFeatureVector::updateOrCreate(
+                ['id' => $annotation->id],
+                [
+                    'job_id' => $annotation->job_id,
+                    'vector' => $row[1],
+                ]
+            );
+        }
+    }
+}