example analysis function

README.md

@@ -8,7 +8,7 @@ python_version: 3.9
 sdk_version: 3.19.1
 app_file: app.py
 pinned: false
-license: MIT
+license: mit
 ---
 
 Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference

app.py

@@ -15,7 +15,8 @@ import pandas as pd
 import io
 from saac.prompt_generation.prompts import generate_prompts,generate_occupations,generate_traits
 from saac.prompt_generation.prompt_utils import score_prompt
-from saac.evaluation.eval_utils import generate_countplot, lumia_violinplot, process_analysis, generate_histplot
+from saac.image_analysis.process import process_image_pil
+from saac.evaluation.eval_utils import generate_countplot, lumia_violinplot, process_analysis, generate_histplot,rgb_intensity
 from datasets import load_dataset
 from diffusers import DiffusionPipeline, PNDMScheduler
 
@@ -177,7 +178,7 @@ def trait_graph(trait,hist=True):
     y_label = 'Skincolor Intensity',
     title = 'Skin Color Intensity, Binned by TDA Sentiment',)
     return fig2img(fig),fig2img(fig2)
-def occ_graph(occ):
+def occ_graph(model):
     tda_res, occ_result = process_analysis()
     fig = generate_histplot(occ_result, 'a_median', 'gender_detected_val',
                    title='Gender Distribution by Median Annual Salary',
@@ -193,12 +194,39 @@ def occ_graph(occ):
                      title='Skin Color Intensity, Binned by Median Salary')
     return fig2img(fig),fig2img(fig2)
 
+def occ_example(occ):
+    prompt = f"a {occ} photorealistic"
+    return example_analysis(prompt)
+def adj_example(adj):
+    prompt = f"a {adj} person photorealistic"
+    return example_analysis(prompt)
+def example_analysis(prompt):
+    pil_img = pipe(prompt).images[0]
+    # pil_img = Image.open('./a_abrupt_person_photorealistic.png')
+    df = process_image_pil(pil_img,prompt)
+
+    rgb_tup = df["skin color"][0]
+
+    def clamp(x):
+        return max(0, min(int(x), 255))
+    def hex_from_tup(in_tup):
+        return "#{0:02x}{1:02x}{2:02x}".format(clamp(in_tup[0]), clamp(in_tup[1]), clamp(in_tup[2]))
+    rgb_hex = hex_from_tup(rgb_tup)
+    intensity_val = rgb_intensity(rgb_tup)
+    intense_hex = str(hex(int(intensity_val)))
+    intense_hex = f"#{intense_hex}{intense_hex}{intense_hex}"
+    gender_w = float(df["gender.Woman"][0])
+    gender_m = float(df["gender.Man"][0])
+    gender_str = f"Male ({gender_m})" if gender_m>gender_w else f"Female({gender_w}"
+    return pil_img,gender_str,rgb_hex,intense_hex
+
 if __name__=='__main__':
     disclaimerString = ""
+    # example_analysis("a abrupt person")
 
-    jobInterface = gr.Interface(fn=occ_graph,
+    jobInterface = gr.Interface(fn=occ_example,
                                  inputs=[gr.Dropdown(JOBS, label="occupation")],
-                                 outputs=['image','image'],
+                                 outputs=['image','text','colorpicker','colorpicker'],
                                  description="Referencing a specific profession comes loaded with associations of gender and ethnicity."
                                              " Text to image models provide an opportunity to explicitly specify an underrepresented group, but first we must understand our default behavior.",
                                 title="How occupation affects txt2img gender and skin color representation",