Skip to content

Fix https://github.com/HiroIshida/detic_ros/issues/82 #2

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
HiroIshida merged 1 commit into from
Jul 18, 2025
Merged

Fix https://github.com/HiroIshida/detic_ros/issues/82 #2

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
8 changes: 5 additions & 3 deletions detic/predictor.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -23,13 +23,15 @@ def get_clip_embeddings(vocabulary, prompt='a '):
cache_file_path = f"/tmp/detic-clip-embeddings-{hash_value}.pt"
if Path(cache_file_path).exists():
print(f"loading embeddings for {vocabulary} from {cache_file_path}")
return torch.load(cache_file_path)
return torch.load(cache_file_path, map_location='cpu')
else:
from detic.modeling.text.text_encoder import build_text_encoder
text_encoder = build_text_encoder(pretrain=True)
text_encoder.eval()
device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')
text_encoder.eval().to(device)
texts = [prompt + x for x in vocabulary]
emb = text_encoder(texts).detach().permute(1, 0).contiguous().cpu()
with torch.no_grad():
emb = text_encoder(texts).detach().permute(1, 0).contiguous().cpu()
print(f"saved embeddings for {vocabulary} to {cache_file_path}")
torch.save(emb, cache_file_path)
return emb
Expand Down