ios/metadata.json

{
 "kind": "video-action-recognition",
 "base_model": "facebook/vjepa2-vitl-fpc16-256-ssv2",
 "bundle": "vjepa2_ssv2_fp16.aimodel",
 "input": {
  "name": "pixel_values_videos",
  "shape": [
   1,
   16,
   3,
   256,
   256
  ],
  "dtype": "float16",
  "preprocess": "frames RGB 0..1 -> (x - mean)/std, mean=[0.485,0.456,0.406], std=[0.229,0.224,0.225], 16 frames uniformly sampled, 256x256 center crop"
 },
 "output": {
  "name": "logits",
  "shape": [
   1,
   174
  ],
  "labels": "labels.json (Something-Something v2)"
 }
}