wip

.gitattributes

@@ -33,3 +33,5 @@ saved_model/**/* 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
+gbif_full_filtered.csv filter=lfs diff=lfs merge=lfs -text
+stl_species.npy filter=lfs diff=lfs merge=lfs -text

app.py

@@ -0,0 +1,83 @@
+import gradio as gr
+import numpy as np
+import matplotlib.pyplot as plt
+from copy import deepcopy
+import pandas as pd
+
+
+def get_index_of_element_containing_word(lst, word):
+    # Create a list of indices where the word is found in the element
+    indices = [i for i, element in enumerate(lst) if word.lower() in element.lower()]
+    # Return the first index found, or -1 if the word is not found in any element
+    return indices[0] if indices else -1
+
+pred_global = None
+
+stl_preds = np.load("stl_species.npy")
+df = pd.read_csv("gbif_full_filtered.csv")
+obs = df.drop_duplicates(subset=["species"])["species"].tolist()
+obs = list(sorted(obs))
+del df
+
+
+def update_fn(val):
+    if val=="Class":
+        return gr.Dropdown(label="Name", choices=class_list, interactive=True)
+    elif val=="Order":
+        return gr.Dropdown(label="Name", choices=order_list, interactive=True)
+    elif val=="Family":
+        return gr.Dropdown(label="Name", choices=family_list, interactive=True)
+    elif val=="Genus":
+        return gr.Dropdown(label="Name", choices=genus_list, interactive=True)
+    elif val=="Species":
+        return gr.Dropdown(label="Name", choices=obs, interactive=True)
+
+def text_fn(taxon, name):
+    global pred_global
+
+    species_index = get_index_of_element_containing_word(obs, name)
+    preds = np.flip(stl_preds[:, species_index].reshape(510, 510), 1)
+
+    pred_global = preds
+    cmap = plt.get_cmap('plasma')
+
+    rgba_img = cmap(preds)
+    rgb_img = np.delete(rgba_img, 3, 2)
+    #return gr.Image(preds, label="Predicted Heatmap", visible=True)
+    return rgb_img
+
+def thresh_fn(val):
+    global pred_global
+    preds = deepcopy(pred_global)
+    preds[preds<val] = 0
+    preds[preds>=val] = 1
+    cmap = plt.get_cmap('plasma')
+
+    rgba_img = cmap(preds)
+    rgb_img = np.delete(rgba_img, 3, 2)
+    return rgb_img
+
+with gr.Blocks() as demo:
+    gr.Markdown(
+    """
+    # Hierarchical Species Distribution Model!
+    This model predicts the distribution of species based on geographic, environmental, and natural language features.
+    """)
+    with gr.Row():
+        inp = gr.Dropdown(label="Taxonomic Hierarchy", choices=["Species"])
+        out = gr.Dropdown(label="Name", interactive=True)
+        inp.change(update_fn, inp, out)
+    
+    with gr.Row():
+        check_button = gr.Button("Run Model")
+    
+    with gr.Row():
+        slider = gr.Slider(minimum=0, maximum=1, step=0.01, default=0.5, label="Confidence Threshold")
+    
+    with gr.Row():
+        pred = gr.Image(label="Predicted Heatmap", visible=True)
+    
+    check_button.click(text_fn, inputs=[inp, out], outputs=[pred])
+    slider.change(thresh_fn, slider, outputs=pred)
+    
+demo.launch()

gbif_full_filtered.csv

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f6e3a753d4444ca34e76402d1081de32412fd8bd6c818747b3a47b244c1ff70c
+size 1402953121

stl_species.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:80be6ce0ac0dc1d11e9644f74aedce5177d66a66d4eb688b33c4b91283e5686c
+size 6212228528