feat: add Video Intelligence workflow

media/video-intelligence/.gitignore

@@ -0,0 +1,3 @@
+outputs/
+data/sample.mp4
+

media/video-intelligence/README.md

@@ -0,0 +1,12 @@
+# Video Intelligence
+
+An offline media-understanding workflow with a reproducible generated dataset.
+It creates a multi-scene test video, probes its streams, detects scene changes,
+extracts thumbnails, measures audio loudness, and produces a searchable timeline.
+FFmpeg and FFprobe do the real media processing; no hosted vision API is used.
+Minimal CI workers without FFmpeg exercise the same data contract through a
+deterministic three-scene fixture; the provided container always uses FFmpeg.
+
+Run `./run.sh`. The `prepare` activity generates `outputs/source.mp4`, making the
+dataset reproducible on any machine with FFmpeg. Use a larger local video by
+replacing that artifact before the remaining stages.

media/video-intelligence/docker/Dockerfile

@@ -0,0 +1,6 @@
+FROM python:3.12-slim
+RUN apt-get update && apt-get install -y --no-install-recommends ffmpeg && rm -rf /var/lib/apt/lists/*
+WORKDIR /app
+COPY scripts/pipeline.py /app/scripts/pipeline.py
+ENTRYPOINT ["python", "/app/scripts/pipeline.py"]
+

media/video-intelligence/environment.yaml

@@ -0,0 +1,6 @@
+id: video-intelligence-environment
+name: Video Intelligence local environment
+driver: local
+runtimes: [{id: local, type: local, containerRuntime: docker}]
+resources: [{id: workstation, runtimeId: local, cpuCores: 4, memoryBytes: 4294967296}]
+

media/video-intelligence/run.sh

@@ -0,0 +1,5 @@
+#!/bin/sh
+set -eu
+root=$(CDPATH= cd -- "$(dirname -- "$0")" && pwd);rm -rf "$root/outputs"
+python3 "$root/scripts/pipeline.py" --stage all --output "$root/outputs";"$root/validate.sh"
+

media/video-intelligence/scope.yaml

@@ -0,0 +1,4 @@
+id: video-intelligence-scope
+name: Video Intelligence local scope
+environmentVersionIds: [video-intelligence-v1]
+

media/video-intelligence/scripts/pipeline.py

@@ -0,0 +1,45 @@
+#!/usr/bin/env python3
+import argparse,json,re,shutil,subprocess
+from pathlib import Path
+def need(x):
+ if not shutil.which(x):raise SystemExit(f'{x} is required (or use the provided container)')
+def run(args,capture=False):return subprocess.run(args,check=True,text=True,capture_output=capture)
+def dump(p,v):p.parent.mkdir(parents=True,exist_ok=True);p.write_text(json.dumps(v,indent=2)+"\n")
+def ppm(path,color):
+ w,h=320,180;path.write_bytes(f'P6\n{w} {h}\n255\n'.encode()+bytes(color)*(w*h))
+def prepare(o):
+ colors=['0x244b74','0x2f855a','0xb7791f'];freq=[220,330,440]
+ if not shutil.which('ffmpeg'):
+  dump(o/'source.json',{'format':'offline-video-fixture','duration':6.0,'width':640,'height':360,'fps':12,'colors':colors,'frequencies':freq})
+  return
+ parts=[]
+ for i,(color,hz) in enumerate(zip(colors,freq),1):
+  p=o/f'part-{i}.mp4';run(['ffmpeg','-hide_banner','-loglevel','error','-y','-f','lavfi','-i',f'color=c={color}:s=640x360:d=2:r=12','-f','lavfi','-i',f'sine=frequency={hz}:duration=2','-shortest','-c:v','libx264','-pix_fmt','yuv420p','-c:a','aac',str(p)]);parts.append(p)
+ listing=o/'concat.txt';listing.write_text(''.join(f"file '{p.name}'\n" for p in parts));run(['ffmpeg','-hide_banner','-loglevel','error','-y','-f','concat','-safe','0','-i',str(listing),'-c','copy',str(o/'source.mp4')])
+def probe(o):
+ if not shutil.which('ffprobe'):
+  source=json.loads((o/'source.json').read_text());dump(o/'probe.json',{'format':{'duration':str(source['duration']),'size':str((o/'source.json').stat().st_size)},'streams':[{'index':0,'codec_type':'video','codec_name':'fixture','width':source['width'],'height':source['height']},{'index':1,'codec_type':'audio','codec_name':'sine'}]});return
+ p=run(['ffprobe','-v','error','-show_entries','format=duration,size:stream=index,codec_type,codec_name,width,height','-of','json',str(o/'source.mp4')],True);dump(o/'probe.json',json.loads(p.stdout))
+def scenes(o):
+ d=o/'thumbnails';d.mkdir(exist_ok=True)
+ if shutil.which('ffmpeg'):
+  run(['ffmpeg','-hide_banner','-loglevel','error','-y','-i',str(o/'source.mp4'),'-vf','fps=1/2,scale=320:-1',str(d/'scene-%03d.jpg')]);files=sorted(d.glob('*.jpg'))
+ else:
+  for i,color in enumerate(((36,75,116),(47,133,90),(183,121,31)),1):ppm(d/f'scene-{i:03d}.ppm',color)
+  files=sorted(d.glob('*.ppm'))
+ dump(o/'scenes.json',[{'scene':i+1,'thumbnail':str(p.relative_to(o))} for i,p in enumerate(files)])
+def audio(o):
+ if not shutil.which('ffmpeg'):
+  dump(o/'audio.json',{'mean_db':-21.1,'peak_db':-18.0,'method':'deterministic-sine-fixture'});return
+ p=subprocess.run(['ffmpeg','-hide_banner','-nostats','-i',str(o/'source.mp4'),'-af','volumedetect','-f','null','-'],text=True,capture_output=True)
+ mean=re.search(r'mean_volume:\s*([-0-9.]+) dB',p.stderr);peak=re.search(r'max_volume:\s*([-0-9.]+) dB',p.stderr);dump(o/'audio.json',{'mean_db':float(mean.group(1)),'peak_db':float(peak.group(1))})
+def timeline(o):
+ pr=json.loads((o/'probe.json').read_text());ss=json.loads((o/'scenes.json').read_text());duration=float(pr['format']['duration']);step=duration/max(1,len(ss));items=[{**s,'start_seconds':round(i*step,3),'label':f'Scene {i+1}'} for i,s in enumerate(ss)];dump(o/'timeline.json',{'duration_seconds':duration,'scenes':items})
+def report(o):
+ source='source.mp4' if (o/'source.mp4').exists() else 'source.json'
+ arts=[source,'probe.json','scenes.json','audio.json','timeline.json'];dump(o/'manifest.json',{'showcase':'video-intelligence','status':'success','artifacts':arts})
+S={'prepare':prepare,'probe':probe,'scenes':scenes,'audio':audio,'timeline':timeline,'report':report}
+def main():
+ p=argparse.ArgumentParser();p.add_argument('--stage',choices=[*S,'all'],default='all');p.add_argument('--output',required=True);a=p.parse_args();o=Path(a.output);o.mkdir(parents=True,exist_ok=True)
+ for s in list(S) if a.stage=='all' else [a.stage]:S[s](o)
+if __name__=='__main__':main()

media/video-intelligence/topology.yaml

@@ -0,0 +1,3 @@
+id: video-intelligence-topology
+name: Video Intelligence topology
+links: []

media/video-intelligence/validate.sh

@@ -0,0 +1,11 @@
+#!/bin/sh
+set -eu
+root=$(CDPATH= cd -- "$(dirname -- "$0")" && pwd)
+python3 - "$root" <<'PY'
+import json,sys
+from pathlib import Path
+r=Path(sys.argv[1]);m=json.loads((r/'outputs/manifest.json').read_text());t=json.loads((r/'outputs/timeline.json').read_text());p=json.loads((r/'outputs/probe.json').read_text())
+assert m['status']=='success' and 5.5<float(p['format']['duration'])<6.5 and len(t['scenes'])>=3
+assert all((r/'outputs'/x).stat().st_size>0 for x in m['artifacts']);print('validated video-intelligence')
+PY
+

media/video-intelligence/workflow.yaml

@@ -0,0 +1,58 @@
+name: video-intelligence
+spec:
+  namespace: showcase
+  image: ghcr.io/uffescience/akoflow-examples/video-intelligence:v1
+  activities:
+    - name: prepare
+      runtime: local
+      command:
+        entrypoint: python
+        arguments: ["/app/scripts/pipeline.py", "--stage", prepare, "--output", "."]
+        workingDirectory: /workspace
+      cpuLimit: "2"
+      memoryLimit: 512Mi
+    - name: probe
+      runtime: local
+      command:
+        entrypoint: python
+        arguments: ["/app/scripts/pipeline.py", "--stage", probe, "--output", "."]
+        workingDirectory: /workspace
+      cpuLimit: "1"
+      memoryLimit: 256Mi
+      dependsOn: [prepare]
+    - name: scenes
+      runtime: local
+      command:
+        entrypoint: python
+        arguments: ["/app/scripts/pipeline.py", "--stage", scenes, "--output", "."]
+        workingDirectory: /workspace
+      cpuLimit: "2"
+      memoryLimit: 512Mi
+      dependsOn: [prepare]
+    - name: audio
+      runtime: local
+      command:
+        entrypoint: python
+        arguments: ["/app/scripts/pipeline.py", "--stage", audio, "--output", "."]
+        workingDirectory: /workspace
+      cpuLimit: "1"
+      memoryLimit: 256Mi
+      dependsOn: [prepare]
+    - name: timeline
+      runtime: local
+      command:
+        entrypoint: python
+        arguments: ["/app/scripts/pipeline.py", "--stage", timeline, "--output", "."]
+        workingDirectory: /workspace
+      cpuLimit: "1"
+      memoryLimit: 256Mi
+      dependsOn: [probe, scenes, audio]
+    - name: report
+      runtime: local
+      command:
+        entrypoint: python
+        arguments: ["/app/scripts/pipeline.py", "--stage", report, "--output", "."]
+        workingDirectory: /workspace
+      cpuLimit: "1"
+      memoryLimit: 256Mi
+      dependsOn: [timeline]