VaianiLorenzo
commited on
Commit
•
b8d9e31
1
Parent(s):
89e64c9
Update app.py
Browse files
app.py
CHANGED
@@ -188,7 +188,7 @@ class CLIPDemo:
|
|
188 |
def compute_image_embeddings(self, image_paths: list):
|
189 |
self.image_paths = image_paths
|
190 |
dataloader = DataLoader(VisionDataset(
|
191 |
-
image_paths=image_paths), batch_size=self.batch_size
|
192 |
embeddings = []
|
193 |
with torch.no_grad():
|
194 |
|
@@ -316,7 +316,7 @@ def draw_audio(
|
|
316 |
vision_encoder = CLIPVisionModel.from_pretrained(CLIP_VISION_MODEL_PATH, local_files_only=True).to(device)
|
317 |
tokenizer = AutoTokenizer.from_pretrained(TEXT_MODEL)
|
318 |
model = CLIPDemo(vision_encoder=vision_encoder, text_encoder=text_encoder, tokenizer=tokenizer, device=device)
|
319 |
-
model.compute_image_embeddings(glob.glob(SPECTROGRAMS_PATH+"/*.jpeg")[:
|
320 |
st.session_state["model"] = model
|
321 |
#st.session_state['model'] = CLIPDemo(vision_encoder=vision_encoder, text_encoder=text_encoder, tokenizer=tokenizer)
|
322 |
#st.session_state.model.compute_image_embeddings(glob.glob("/data1/mlaquatra/TSOAI_hack/data/spectrograms/*.jpeg")[:100])
|
@@ -384,7 +384,7 @@ def draw_camera(
|
|
384 |
vision_encoder = CLIPVisionModel.from_pretrained(CLIP_VISION_MODEL_PATH, local_files_only=True).to(device)
|
385 |
tokenizer = AutoTokenizer.from_pretrained(TEXT_MODEL)
|
386 |
model = CLIPDemo(vision_encoder=vision_encoder, text_encoder=text_encoder, tokenizer=tokenizer, device=device)
|
387 |
-
model.compute_image_embeddings(glob.glob(SPECTROGRAMS_PATH + "/*.jpeg")[:
|
388 |
st.session_state["model"] = model
|
389 |
#st.session_state['model'] = CLIPDemo(vision_encoder=vision_encoder, text_encoder=text_encoder, tokenizer=tokenizer)
|
390 |
#st.session_state.model.compute_image_embeddings(glob.glob("/data1/mlaquatra/TSOAI_hack/data/spectrograms/*.jpeg")[:100])
|
|
|
188 |
def compute_image_embeddings(self, image_paths: list):
|
189 |
self.image_paths = image_paths
|
190 |
dataloader = DataLoader(VisionDataset(
|
191 |
+
image_paths=image_paths), batch_size=self.batch_size)
|
192 |
embeddings = []
|
193 |
with torch.no_grad():
|
194 |
|
|
|
316 |
vision_encoder = CLIPVisionModel.from_pretrained(CLIP_VISION_MODEL_PATH, local_files_only=True).to(device)
|
317 |
tokenizer = AutoTokenizer.from_pretrained(TEXT_MODEL)
|
318 |
model = CLIPDemo(vision_encoder=vision_encoder, text_encoder=text_encoder, tokenizer=tokenizer, device=device)
|
319 |
+
model.compute_image_embeddings(glob.glob(SPECTROGRAMS_PATH+"/*.jpeg")[:1000])
|
320 |
st.session_state["model"] = model
|
321 |
#st.session_state['model'] = CLIPDemo(vision_encoder=vision_encoder, text_encoder=text_encoder, tokenizer=tokenizer)
|
322 |
#st.session_state.model.compute_image_embeddings(glob.glob("/data1/mlaquatra/TSOAI_hack/data/spectrograms/*.jpeg")[:100])
|
|
|
384 |
vision_encoder = CLIPVisionModel.from_pretrained(CLIP_VISION_MODEL_PATH, local_files_only=True).to(device)
|
385 |
tokenizer = AutoTokenizer.from_pretrained(TEXT_MODEL)
|
386 |
model = CLIPDemo(vision_encoder=vision_encoder, text_encoder=text_encoder, tokenizer=tokenizer, device=device)
|
387 |
+
model.compute_image_embeddings(glob.glob(SPECTROGRAMS_PATH + "/*.jpeg")[:1000])
|
388 |
st.session_state["model"] = model
|
389 |
#st.session_state['model'] = CLIPDemo(vision_encoder=vision_encoder, text_encoder=text_encoder, tokenizer=tokenizer)
|
390 |
#st.session_state.model.compute_image_embeddings(glob.glob("/data1/mlaquatra/TSOAI_hack/data/spectrograms/*.jpeg")[:100])
|