cudanexus/detic 🖼️❓📝 → ❓

▶️ 307 runs 📅 Dec 2023 ⚙️ Cog 0.8.6 🔗 GitHub ⚖️ License
image-object-detection open-vocabulary

About

Detecting Twenty-thousand Classes using Image-level Supervision

Example Output

Output

{"image":"https://replicate.delivery/pbxt/8Z3ePZdM4YQGCCENdN8egqBDXAsCDPRIkczkeKA6QntNtwLkA/out.png","jsona":"https://replicate.delivery/pbxt/1tYTuPjeYO2nVS0Ll1hDIMIc2524f3oDAUlNb9E9pWbnW4FSA/output.json"}

Performance Metrics

2.60s Prediction Time
120.88s Total Time
All Input Parameters 
{
  "image": "https://replicate.delivery/pbxt/K6zgH3un6xi4l07DFSUOgHKzOIbWktlvmc348VZbRDqSbqeV/k67kjlB.jpeg",
  "vocabulary": "lvis",
  "custom_vocabulary": "None"
}
Input Parameters
image (required) Type: string
input image
vocabulary Default: lvis
Choose vocabulary
custom_vocabulary (required) Type: string
Type your own vocabularies, separated by coma ','
Output Schema
image Type: stringFormat: uri
Image
jsona Type: stringFormat: uri
Jsona
Example Execution Logs 
Resetting zs_weight datasets/metadata/lvis_v1_clip_a+cname.npy
JSON data written to /tmp/tmpe847452h/output.json
Version Details
Version ID
64998fbfaa64b44ef350920eeb18c8db1cda7dbcea9945e3996fefda407815bf
Version Created
January 3, 2024
Run on Replicate →