initial

maskformer_swin_large_coco/app.py

@@ -0,0 +1,43 @@
+import gradio as gr
+from transformers import MaskFormerImageProcessor, MaskFormerForInstanceSegmentation
+from PIL import Image
+def inference(img):
+    # load MaskFormer fine-tuned on COCO panoptic segmentation
+    model_path = "maskformer-swin-large-coco"
+    processor = MaskFormerImageProcessor.from_pretrained(model_path)
+    model = MaskFormerForInstanceSegmentation.from_pretrained(model_path)
+
+    inputs = processor(images=img, return_tensors="pt")
+
+    outputs = model(**inputs)
+    # model predicts class_queries_logits of shape `(batch_size, num_queries)`
+    # and masks_queries_logits of shape `(batch_size, num_queries, height, width)`
+    class_queries_logits = outputs.class_queries_logits
+    masks_queries_logits = outputs.masks_queries_logits
+
+    # you can pass them to processor for postprocessing
+    result = processor.post_process_panoptic_segmentation(outputs, target_sizes=[img.size[::-1]])[0]
+    # we refer to the demo notebooks for visualization (see "Resources" section in the MaskFormer docs)
+    predicted_panoptic_map = result["segmentation"]
+    return predicted_panoptic_map
+
+examples=[['example_cat.jpg']]
+
+with gr.Blocks() as demo:
+    gr.Markdown(
+    """
+    # Panorama segmentation:maskformer-swin-large-coco
+    Gradio Demo for maskformer-swin-large-coco. To use it, simply upload your image, or click one of the examples to load them.
+    """)
+    with gr.Row():
+        image_input = gr.Image(type="pil")
+        text_output = gr.Textbox()
+    image_button = gr.Button("上传")
+    image_button.click(inference, inputs=image_input, outputs=text_output)
+    gr.Examples(examples,inputs=image_input)
+
+demo.launch()
+
+
+
+

maskformer_swin_large_coco/maskformer-swin-large-coco

@@ -0,0 +1 @@
+Subproject commit 5f009df2186c5841582ecd419409b56041180144