Simon commited on
Commit
65a6efd
·
1 Parent(s): 8217c57
Files changed (1) hide show
  1. app.py +3 -3
app.py CHANGED
@@ -195,9 +195,9 @@ with gr.Blocks() as demo:
195
 
196
  # Context pulled directly from the paper
197
  gr.Markdown("""
198
- * **Semantic Match (Idea-Expression Dichotomy):** Evaluates if works share an underlying conceptual basis. Models like CLIP link visual data to semantic concepts to bridge the "semantic gap"[cite: 1]. This acts as an initial threshold, as copyright protects concrete expression rather than mere abstract ideas[cite: 1].
199
- * **Structural Match (Substantial Similarity):** Detects structural relationships through spatial mapping[cite: 1]. This addresses how the specific combination and arrangement of individual elements can form part of the protected expression[cite: 1].
200
- * **Patch Match (Fragmented Literal Similarity):** Identifies "scattered literal copying" where specific fragments of a protected work are reproduced directly or in a highly proximate manner, without the new work as a whole having to resemble the original[cite: 1]. Localized patch-based matching algorithms divide the image to detect these copied fragments regardless of where they are placed in the new works[cite: 1].
201
  """)
202
 
203
  with gr.Row():
 
195
 
196
  # Context pulled directly from the paper
197
  gr.Markdown("""
198
+ * **Semantic Match (Idea-Expression Dichotomy):** Evaluates if works share an underlying conceptual basis. Models like CLIP link visual data to semantic concepts to bridge the "semantic gap". This acts as an initial threshold, as copyright protects concrete expression rather than mere abstract ideas.
199
+ * **Structural Match (Substantial Similarity):** Detects structural relationships through spatial mapping. This addresses how the specific combination and arrangement of individual elements can form part of the protected expression.
200
+ * **Patch Match (Fragmented Literal Similarity):** Identifies "scattered literal copying" where specific fragments of a protected work are reproduced directly or in a highly proximate manner, without the new work as a whole having to resemble the original. Localized patch-based matching algorithms divide the image to detect these copied fragments regardless of where they are placed in the new works.
201
  """)
202
 
203
  with gr.Row():