localhorst/CLIP_on_Tesla_K20Xm
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

label and user text

Browse Source
This commit is contained in:
AK391 2021-03-23 13:33:19 -04:00
parent 6ca785bda0
commit e8945f5adb
1 changed files with 14 additions and 10 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

24
demo.py
View File

@ -6,14 +6,16 @@ import gradio as gr
# Load the model # Load the model
device = "cuda" if torch.cuda.is_available() else "cpu" device = "cuda" if torch.cuda.is_available() else "cpu"
model, transform = clip.load("ViT-B/32", device=device) model, preprocess = clip.load('ViT-B/32', device)
# Download the dataset # Download the dataset
cifar100 = CIFAR100(root=os.path.expanduser("~/.cache"), download=True, train=False) cifar100 = CIFAR100(root=os.path.expanduser("~/.cache"), download=True, train=False)
def classify(img): def classify(img, user_text):
image = transform(img).unsqueeze(0).to(device) image = preprocess(img).unsqueeze(0).to(device)
text_inputs = torch.cat([clip.tokenize(f"a photo of a {c}") for c in cifar100.classes]).to(device) user_texts = user_text.split(",")
text_sources = cifar100.classes + user_texts
text_inputs = torch.cat([clip.tokenize(f"a photo of a {c}") for c in text_sources]).to(device)
# Calculate features # Calculate features
with torch.no_grad(): with torch.no_grad():
@ -26,14 +28,16 @@ def classify(img):
similarity = (100.0 * image_features @ text_features.T).softmax(dim=-1) similarity = (100.0 * image_features @ text_features.T).softmax(dim=-1)
values, indices = similarity[0].topk(5) values, indices = similarity[0].topk(5)
text="" result = {}
# Print the result
for value, index in zip(values, indices): for value, index in zip(values, indices):
text+=f"{cifar100.classes[index]:>16s}: {100 * value.item():.2f}%\n" result[text_sources[index]] = value.item()
return text return result
inputs = gr.inputs.Image(type='pil', label="Original Image") inputs = [
outputs = gr.outputs.Textbox(type="str", label="Text Output") gr.inputs.Image(type='pil', label="Original Image"),
gr.inputs.Textbox(lines=1)
]
outputs = gr.outputs.Label(type="confidences",num_top_classes=5)
title = "CLIP" title = "CLIP"
description = "CLIP demo" description = "CLIP demo"