Duplicate from nielsr/comparing-captioning-models

Co-authored-by: Niels Rogge <nielsr@users.noreply.huggingface.co>

.gitattributes

@@ -0,0 +1,34 @@
+*.7z filter=lfs diff=lfs merge=lfs -text
+*.arrow filter=lfs diff=lfs merge=lfs -text
+*.bin filter=lfs diff=lfs merge=lfs -text
+*.bz2 filter=lfs diff=lfs merge=lfs -text
+*.ckpt filter=lfs diff=lfs merge=lfs -text
+*.ftz filter=lfs diff=lfs merge=lfs -text
+*.gz filter=lfs diff=lfs merge=lfs -text
+*.h5 filter=lfs diff=lfs merge=lfs -text
+*.joblib filter=lfs diff=lfs merge=lfs -text
+*.lfs.* filter=lfs diff=lfs merge=lfs -text
+*.mlmodel filter=lfs diff=lfs merge=lfs -text
+*.model filter=lfs diff=lfs merge=lfs -text
+*.msgpack filter=lfs diff=lfs merge=lfs -text
+*.npy filter=lfs diff=lfs merge=lfs -text
+*.npz filter=lfs diff=lfs merge=lfs -text
+*.onnx filter=lfs diff=lfs merge=lfs -text
+*.ot filter=lfs diff=lfs merge=lfs -text
+*.parquet filter=lfs diff=lfs merge=lfs -text
+*.pb filter=lfs diff=lfs merge=lfs -text
+*.pickle filter=lfs diff=lfs merge=lfs -text
+*.pkl filter=lfs diff=lfs merge=lfs -text
+*.pt filter=lfs diff=lfs merge=lfs -text
+*.pth filter=lfs diff=lfs merge=lfs -text
+*.rar filter=lfs diff=lfs merge=lfs -text
+*.safetensors filter=lfs diff=lfs merge=lfs -text
+saved_model/**/* filter=lfs diff=lfs merge=lfs -text
+*.tar.* filter=lfs diff=lfs merge=lfs -text
+*.tflite filter=lfs diff=lfs merge=lfs -text
+*.tgz filter=lfs diff=lfs merge=lfs -text
+*.wasm filter=lfs diff=lfs merge=lfs -text
+*.xz filter=lfs diff=lfs merge=lfs -text
+*.zip filter=lfs diff=lfs merge=lfs -text
+*.zst filter=lfs diff=lfs merge=lfs -text
+*tfevents* filter=lfs diff=lfs merge=lfs -text

README.md

@@ -0,0 +1,13 @@
+---
+title: Comparing Captioning Models
+emoji: 🔥
+colorFrom: yellow
+colorTo: pink
+sdk: gradio
+sdk_version: 3.15.0
+app_file: app.py
+pinned: false
+duplicated_from: nielsr/comparing-captioning-models
+---
+
+Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference

app.py

@@ -0,0 +1,105 @@
+import gradio as gr
+from transformers import AutoProcessor, AutoTokenizer, AutoImageProcessor, AutoModelForCausalLM, BlipForConditionalGeneration, Blip2ForConditionalGeneration, VisionEncoderDecoderModel
+import torch
+import open_clip
+
+from huggingface_hub import hf_hub_download
+
+torch.hub.download_url_to_file('http://images.cocodataset.org/val2017/000000039769.jpg', 'cats.jpg')
+torch.hub.download_url_to_file('https://huggingface.co/datasets/nielsr/textcaps-sample/resolve/main/stop_sign.png', 'stop_sign.png')
+torch.hub.download_url_to_file('https://cdn.openai.com/dall-e-2/demos/text2im/astronaut/horse/photo/0.jpg', 'astronaut.jpg')
+
+# git_processor_base = AutoProcessor.from_pretrained("microsoft/git-base-coco")
+# git_model_base = AutoModelForCausalLM.from_pretrained("microsoft/git-base-coco")
+
+git_processor_large_coco = AutoProcessor.from_pretrained("microsoft/git-large-coco")
+git_model_large_coco = AutoModelForCausalLM.from_pretrained("microsoft/git-large-coco")
+
+git_processor_large_textcaps = AutoProcessor.from_pretrained("microsoft/git-large-r-textcaps")
+git_model_large_textcaps = AutoModelForCausalLM.from_pretrained("microsoft/git-large-r-textcaps")
+
+# blip_processor_base = AutoProcessor.from_pretrained("Salesforce/blip-image-captioning-base")
+# blip_model_base = BlipForConditionalGeneration.from_pretrained("Salesforce/blip-image-captioning-base")
+
+blip_processor_large = AutoProcessor.from_pretrained("Salesforce/blip-image-captioning-large")
+blip_model_large = BlipForConditionalGeneration.from_pretrained("Salesforce/blip-image-captioning-large")
+
+blip2_processor = AutoProcessor.from_pretrained("Salesforce/blip2-opt-2.7b")
+blip2_model = Blip2ForConditionalGeneration.from_pretrained("Salesforce/blip2-opt-2.7b")
+
+# vitgpt_processor = AutoImageProcessor.from_pretrained("nlpconnect/vit-gpt2-image-captioning")
+# vitgpt_model = VisionEncoderDecoderModel.from_pretrained("nlpconnect/vit-gpt2-image-captioning")
+# vitgpt_tokenizer = AutoTokenizer.from_pretrained("nlpconnect/vit-gpt2-image-captioning")
+
+coca_model, _, coca_transform = open_clip.create_model_and_transforms(
+  model_name="coca_ViT-L-14",
+  pretrained="mscoco_finetuned_laion2B-s13B-b90k"
+)
+
+device = "cuda" if torch.cuda.is_available() else "cpu"
+
+# git_model_base.to(device)
+# blip_model_base.to(device)
+git_model_large_coco.to(device)
+git_model_large_textcaps.to(device)
+blip_model_large.to(device)
+# vitgpt_model.to(device)
+coca_model.to(device)
+blip2_model.to(device)
+
+def generate_caption(processor, model, image, tokenizer=None):
+    inputs = processor(images=image, return_tensors="pt").to(device)
+    
+    generated_ids = model.generate(pixel_values=inputs.pixel_values, max_length=50)
+
+    if tokenizer is not None:
+        generated_caption = tokenizer.batch_decode(generated_ids, skip_special_tokens=True)[0]
+    else:
+        generated_caption = processor.batch_decode(generated_ids, skip_special_tokens=True)[0]
+   
+    return generated_caption
+
+
+def generate_caption_coca(model, transform, image):
+    im = transform(image).unsqueeze(0).to(device)
+    with torch.no_grad(), torch.cuda.amp.autocast():
+        generated = model.generate(im, seq_len=20)
+    return open_clip.decode(generated[0].detach()).split("<end_of_text>")[0].replace("<start_of_text>", "")
+
+
+def generate_captions(image):
+    # caption_git_base = generate_caption(git_processor_base, git_model_base, image)
+
+    caption_git_large_coco = generate_caption(git_processor_large_coco, git_model_large_coco, image)
+
+    caption_git_large_textcaps = generate_caption(git_processor_large_textcaps, git_model_large_textcaps, image)
+
+    # caption_blip_base = generate_caption(blip_processor_base, blip_model_base, image)
+
+    caption_blip_large = generate_caption(blip_processor_large, blip_model_large, image)
+
+    # caption_vitgpt = generate_caption(vitgpt_processor, vitgpt_model, image, vitgpt_tokenizer)
+
+    caption_coca = generate_caption_coca(coca_model, coca_transform, image)
+
+    caption_blip2 = generate_caption(blip2_processor, blip2_model, image)
+
+    return caption_git_large_coco, caption_git_large_textcaps, caption_blip_large, caption_coca, caption_blip2
+
+   
+examples = [["cats.jpg"], ["stop_sign.png"], ["astronaut.jpg"]]
+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 2.7b")] 
+
+title = "Interactive demo: comparing image captioning models"
+description = "Gradio Demo to compare GIT, BLIP, CoCa, and BLIP-2, 4 state-of-the-art vision+language models. To use it, simply upload your image and click 'submit', or click one of the examples to load them. Read more at the links below."
+article = "<p style='text-align: center'><a href='https://huggingface.co/docs/transformers/main/model_doc/blip' target='_blank'>BLIP docs</a> | <a href='https://huggingface.co/docs/transformers/main/model_doc/git' target='_blank'>GIT docs</a></p>"
+
+interface = gr.Interface(fn=generate_captions, 
+                         inputs=gr.inputs.Image(type="pil"),
+                         outputs=outputs,
+                         examples=examples, 
+                         title=title,
+                         description=description,
+                         article=article, 
+                         enable_queue=True)
+interface.launch(debug=True)

requirements.txt

@@ -0,0 +1,3 @@
+git+https://github.com/huggingface/transformers.git@main
+torch
+open_clip_torch