enabled just git large
Browse files
app.py
CHANGED
@@ -11,17 +11,17 @@ torch.hub.download_url_to_file('http://images.cocodataset.org/val2017/0000000397
|
|
11 |
torch.hub.download_url_to_file('https://huggingface.co/datasets/nielsr/textcaps-sample/resolve/main/stop_sign.png', 'stop_sign.png')
|
12 |
torch.hub.download_url_to_file('https://cdn.openai.com/dall-e-2/demos/text2im/astronaut/horse/photo/0.jpg', 'astronaut.jpg')
|
13 |
|
14 |
-
git_processor_base = AutoProcessor.from_pretrained("microsoft/git-base-coco")
|
15 |
-
git_model_base = AutoModelForCausalLM.from_pretrained("microsoft/git-base-coco")
|
16 |
|
17 |
-
|
18 |
-
|
19 |
|
20 |
# git_processor_large_textcaps = AutoProcessor.from_pretrained("microsoft/git-large-r-textcaps")
|
21 |
# git_model_large_textcaps = AutoModelForCausalLM.from_pretrained("microsoft/git-large-r-textcaps")
|
22 |
|
23 |
-
blip_processor_base = AutoProcessor.from_pretrained("Salesforce/blip-image-captioning-base")
|
24 |
-
blip_model_base = BlipForConditionalGeneration.from_pretrained("Salesforce/blip-image-captioning-base")
|
25 |
|
26 |
# blip_processor_large = AutoProcessor.from_pretrained("Salesforce/blip-image-captioning-large")
|
27 |
# blip_model_large = BlipForConditionalGeneration.from_pretrained("Salesforce/blip-image-captioning-large")
|
@@ -43,9 +43,9 @@ blip_model_base = BlipForConditionalGeneration.from_pretrained("Salesforce/blip-
|
|
43 |
|
44 |
device = "cuda" if torch.cuda.is_available() else "cpu"
|
45 |
|
46 |
-
git_model_base.to(device)
|
47 |
-
blip_model_base.to(device)
|
48 |
-
|
49 |
# git_model_large_textcaps.to(device)
|
50 |
# blip_model_large.to(device)
|
51 |
# vitgpt_model.to(device)
|
@@ -76,13 +76,13 @@ def generate_caption_coca(model, transform, image):
|
|
76 |
|
77 |
|
78 |
def generate_captions(image):
|
79 |
-
caption_git_base = generate_caption(git_processor_base, git_model_base, image)
|
80 |
|
81 |
-
|
82 |
|
83 |
# caption_git_large_textcaps = generate_caption(git_processor_large_textcaps, git_model_large_textcaps, image)
|
84 |
|
85 |
-
caption_blip_base = generate_caption(blip_processor_base, blip_model_base, image)
|
86 |
|
87 |
# caption_blip_large = generate_caption(blip_processor_large, blip_model_large, image)
|
88 |
|
@@ -95,17 +95,17 @@ def generate_captions(image):
|
|
95 |
# caption_blip2_8_bit = generate_caption(blip2_processor_8_bit, blip2_model_8_bit, image, use_float_16=True).strip()
|
96 |
|
97 |
# return caption_git_large_coco, caption_git_large_textcaps, caption_blip_large, caption_coca, caption_blip2_8_bit
|
98 |
-
return
|
99 |
|
100 |
|
101 |
|
102 |
examples = [["cats.jpg"], ["stop_sign.png"], ["astronaut.jpg"]]
|
103 |
# outputs = [gr.outputs.Textbox(label="Caption generated by GIT-large fine-tuned on COCO"), gr.outputs.Textbox(label="Caption generated by GIT-large fine-tuned on TextCaps"), gr.outputs.Textbox(label="Caption generated by BLIP-large"), gr.outputs.Textbox(label="Caption generated by CoCa"), gr.outputs.Textbox(label="Caption generated by BLIP-2 OPT 6.7b")]
|
104 |
outputs = [
|
105 |
-
gr.outputs.Textbox(label="Caption generated by GIT-
|
106 |
# gr.outputs.Textbox(label="Caption generated by GIT-large fine-tuned on COCO"),
|
107 |
# gr.outputs.Textbox(label="Caption generated by GIT-large fine-tuned on TextCaps"),
|
108 |
-
gr.outputs.Textbox(label="Caption generated by BLIP-base"),
|
109 |
# gr.outputs.Textbox(label="Caption generated by BLIP-large"),
|
110 |
# gr.outputs.Textbox(label="Caption generated by vitgpt")
|
111 |
]
|
|
|
11 |
torch.hub.download_url_to_file('https://huggingface.co/datasets/nielsr/textcaps-sample/resolve/main/stop_sign.png', 'stop_sign.png')
|
12 |
torch.hub.download_url_to_file('https://cdn.openai.com/dall-e-2/demos/text2im/astronaut/horse/photo/0.jpg', 'astronaut.jpg')
|
13 |
|
14 |
+
# git_processor_base = AutoProcessor.from_pretrained("microsoft/git-base-coco")
|
15 |
+
# git_model_base = AutoModelForCausalLM.from_pretrained("microsoft/git-base-coco")
|
16 |
|
17 |
+
git_processor_large_coco = AutoProcessor.from_pretrained("microsoft/git-large-coco")
|
18 |
+
git_model_large_coco = AutoModelForCausalLM.from_pretrained("microsoft/git-large-coco")
|
19 |
|
20 |
# git_processor_large_textcaps = AutoProcessor.from_pretrained("microsoft/git-large-r-textcaps")
|
21 |
# git_model_large_textcaps = AutoModelForCausalLM.from_pretrained("microsoft/git-large-r-textcaps")
|
22 |
|
23 |
+
# blip_processor_base = AutoProcessor.from_pretrained("Salesforce/blip-image-captioning-base")
|
24 |
+
# blip_model_base = BlipForConditionalGeneration.from_pretrained("Salesforce/blip-image-captioning-base")
|
25 |
|
26 |
# blip_processor_large = AutoProcessor.from_pretrained("Salesforce/blip-image-captioning-large")
|
27 |
# blip_model_large = BlipForConditionalGeneration.from_pretrained("Salesforce/blip-image-captioning-large")
|
|
|
43 |
|
44 |
device = "cuda" if torch.cuda.is_available() else "cpu"
|
45 |
|
46 |
+
# git_model_base.to(device)
|
47 |
+
# blip_model_base.to(device)
|
48 |
+
git_model_large_coco.to(device)
|
49 |
# git_model_large_textcaps.to(device)
|
50 |
# blip_model_large.to(device)
|
51 |
# vitgpt_model.to(device)
|
|
|
76 |
|
77 |
|
78 |
def generate_captions(image):
|
79 |
+
# caption_git_base = generate_caption(git_processor_base, git_model_base, image)
|
80 |
|
81 |
+
caption_git_large_coco = generate_caption(git_processor_large_coco, git_model_large_coco, image)
|
82 |
|
83 |
# caption_git_large_textcaps = generate_caption(git_processor_large_textcaps, git_model_large_textcaps, image)
|
84 |
|
85 |
+
# caption_blip_base = generate_caption(blip_processor_base, blip_model_base, image)
|
86 |
|
87 |
# caption_blip_large = generate_caption(blip_processor_large, blip_model_large, image)
|
88 |
|
|
|
95 |
# caption_blip2_8_bit = generate_caption(blip2_processor_8_bit, blip2_model_8_bit, image, use_float_16=True).strip()
|
96 |
|
97 |
# return caption_git_large_coco, caption_git_large_textcaps, caption_blip_large, caption_coca, caption_blip2_8_bit
|
98 |
+
return caption_git_large_coco
|
99 |
|
100 |
|
101 |
|
102 |
examples = [["cats.jpg"], ["stop_sign.png"], ["astronaut.jpg"]]
|
103 |
# outputs = [gr.outputs.Textbox(label="Caption generated by GIT-large fine-tuned on COCO"), gr.outputs.Textbox(label="Caption generated by GIT-large fine-tuned on TextCaps"), gr.outputs.Textbox(label="Caption generated by BLIP-large"), gr.outputs.Textbox(label="Caption generated by CoCa"), gr.outputs.Textbox(label="Caption generated by BLIP-2 OPT 6.7b")]
|
104 |
outputs = [
|
105 |
+
gr.outputs.Textbox(label="Caption generated by GIT-large fine-tuned on COCO"),
|
106 |
# gr.outputs.Textbox(label="Caption generated by GIT-large fine-tuned on COCO"),
|
107 |
# gr.outputs.Textbox(label="Caption generated by GIT-large fine-tuned on TextCaps"),
|
108 |
+
# gr.outputs.Textbox(label="Caption generated by BLIP-base"),
|
109 |
# gr.outputs.Textbox(label="Caption generated by BLIP-large"),
|
110 |
# gr.outputs.Textbox(label="Caption generated by vitgpt")
|
111 |
]
|