Spaces:
Running
on
Zero
Running
on
Zero
Update app.py
Browse files
app.py
CHANGED
|
@@ -121,34 +121,34 @@ def process_image(image, task_prompt, text_input=None):
|
|
| 121 |
elif task_prompt == '<OD>':
|
| 122 |
results = run_example(task_prompt, image)
|
| 123 |
fig = plot_bbox(image, results['<OD>'])
|
| 124 |
-
return
|
| 125 |
elif task_prompt == '<DENSE_REGION_CAPTION>':
|
| 126 |
results = run_example(task_prompt, image)
|
| 127 |
fig = plot_bbox(image, results['<DENSE_REGION_CAPTION>'])
|
| 128 |
-
return
|
| 129 |
elif task_prompt == '<REGION_PROPOSAL>':
|
| 130 |
results = run_example(task_prompt, image)
|
| 131 |
fig = plot_bbox(image, results['<REGION_PROPOSAL>'])
|
| 132 |
-
return
|
| 133 |
elif task_prompt == '<CAPTION_TO_PHRASE_GROUNDING>':
|
| 134 |
results = run_example(task_prompt, image, text_input)
|
| 135 |
fig = plot_bbox(image, results['<CAPTION_TO_PHRASE_GROUNDING>'])
|
| 136 |
-
return
|
| 137 |
elif task_prompt == '<REFERRING_EXPRESSION_SEGMENTATION>':
|
| 138 |
results = run_example(task_prompt, image, text_input)
|
| 139 |
output_image = copy.deepcopy(image)
|
| 140 |
output_image = draw_polygons(output_image, results['<REFERRING_EXPRESSION_SEGMENTATION>'], fill_mask=True)
|
| 141 |
-
return
|
| 142 |
elif task_prompt == '<REGION_TO_SEGMENTATION>':
|
| 143 |
results = run_example(task_prompt, image, text_input)
|
| 144 |
output_image = copy.deepcopy(image)
|
| 145 |
output_image = draw_polygons(output_image, results['<REGION_TO_SEGMENTATION>'], fill_mask=True)
|
| 146 |
-
return
|
| 147 |
elif task_prompt == '<OPEN_VOCABULARY_DETECTION>':
|
| 148 |
results = run_example(task_prompt, image, text_input)
|
| 149 |
bbox_results = convert_to_od_format(results['<OPEN_VOCABULARY_DETECTION>'])
|
| 150 |
fig = plot_bbox(image, bbox_results)
|
| 151 |
-
return
|
| 152 |
elif task_prompt == '<REGION_TO_CATEGORY>':
|
| 153 |
results = run_example(task_prompt, image, text_input)
|
| 154 |
return results, None
|
|
@@ -162,7 +162,7 @@ def process_image(image, task_prompt, text_input=None):
|
|
| 162 |
results = run_example(task_prompt, image)
|
| 163 |
output_image = copy.deepcopy(image)
|
| 164 |
output_image = draw_ocr_bboxes(output_image, results['<OCR_WITH_REGION>'])
|
| 165 |
-
return
|
| 166 |
else:
|
| 167 |
return "", None # Return empty string and None for unknown task prompts
|
| 168 |
|
|
|
|
| 121 |
elif task_prompt == '<OD>':
|
| 122 |
results = run_example(task_prompt, image)
|
| 123 |
fig = plot_bbox(image, results['<OD>'])
|
| 124 |
+
return results, fig_to_pil(fig)
|
| 125 |
elif task_prompt == '<DENSE_REGION_CAPTION>':
|
| 126 |
results = run_example(task_prompt, image)
|
| 127 |
fig = plot_bbox(image, results['<DENSE_REGION_CAPTION>'])
|
| 128 |
+
return results, fig_to_pil(fig)
|
| 129 |
elif task_prompt == '<REGION_PROPOSAL>':
|
| 130 |
results = run_example(task_prompt, image)
|
| 131 |
fig = plot_bbox(image, results['<REGION_PROPOSAL>'])
|
| 132 |
+
return results, fig_to_pil(fig)
|
| 133 |
elif task_prompt == '<CAPTION_TO_PHRASE_GROUNDING>':
|
| 134 |
results = run_example(task_prompt, image, text_input)
|
| 135 |
fig = plot_bbox(image, results['<CAPTION_TO_PHRASE_GROUNDING>'])
|
| 136 |
+
return results, fig_to_pil(fig)
|
| 137 |
elif task_prompt == '<REFERRING_EXPRESSION_SEGMENTATION>':
|
| 138 |
results = run_example(task_prompt, image, text_input)
|
| 139 |
output_image = copy.deepcopy(image)
|
| 140 |
output_image = draw_polygons(output_image, results['<REFERRING_EXPRESSION_SEGMENTATION>'], fill_mask=True)
|
| 141 |
+
return results, output_image
|
| 142 |
elif task_prompt == '<REGION_TO_SEGMENTATION>':
|
| 143 |
results = run_example(task_prompt, image, text_input)
|
| 144 |
output_image = copy.deepcopy(image)
|
| 145 |
output_image = draw_polygons(output_image, results['<REGION_TO_SEGMENTATION>'], fill_mask=True)
|
| 146 |
+
return results, output_image
|
| 147 |
elif task_prompt == '<OPEN_VOCABULARY_DETECTION>':
|
| 148 |
results = run_example(task_prompt, image, text_input)
|
| 149 |
bbox_results = convert_to_od_format(results['<OPEN_VOCABULARY_DETECTION>'])
|
| 150 |
fig = plot_bbox(image, bbox_results)
|
| 151 |
+
return results, fig_to_pil(fig)
|
| 152 |
elif task_prompt == '<REGION_TO_CATEGORY>':
|
| 153 |
results = run_example(task_prompt, image, text_input)
|
| 154 |
return results, None
|
|
|
|
| 162 |
results = run_example(task_prompt, image)
|
| 163 |
output_image = copy.deepcopy(image)
|
| 164 |
output_image = draw_ocr_bboxes(output_image, results['<OCR_WITH_REGION>'])
|
| 165 |
+
return results, output_image
|
| 166 |
else:
|
| 167 |
return "", None # Return empty string and None for unknown task prompts
|
| 168 |
|