MosquitoCNN / app.py
Sriram Elango
Add application file
ecb645e
raw
history blame
1.12 kB
import gradio as gr
import os
import torch
from PIL import Image
#subprocess.run(["mv","content/custom_data.yaml","./yolov5/data"])
def load_model():
'''
Loading hub model & setting the preferences for the model
'''
model = torch.hub.load('ultralytics/yolov5', 'custom', path='Content/cnn.pt')
model.conf = 0.38
model.dnn=True
model.agnostic=True
return model
model=load_model()
#, force_reload=True
def detect(inp):
#g = (size / max(inp.size)) #gain
#im = im.resize((int(x * g) for x in im.size), Image.ANTIALIAS) # resize
results = model(inp,size=640) # inference
print(results)
results.render() # updates results.imgs with boxes and labels
return Image.fromarray(results.imgs[0])
inp = gr.inputs.Image(type="pil", label="Original Image")
output = gr.outputs.Image(type="pil", label="Output Image")
io=gr.Interface(fn=detect, inputs=inp, outputs=output, title='Mosquito Habitat Identification',theme='peach')
io.launch(debug=True,share=False)
#examples=['Content/4.jpg','Content/10.jpg','Content/18.jpg']