[Python] Waste classifier using confidence rate python [closed]

from transformers import AutoProcessor, AutoModelForImageClassification
from PIL import Image
import torch
import torch.nn.functional as F

processor=AutoProcessor.from_pretrained("watersplash/waste-classification")
model = AutoModelForImageClassification.from_pretrained("watersplash/waste-classification")

WASTE_INFO = {
"organic": {
"harm": "Can release methane if not composted.",
"handling": "Compost or use municipal organic waste bins."
},
"plastic": {
"harm": "Takes centuries to degrade and pollutes oceans.",
"handling": "Recycle based on plastic code or use proper disposal services."
},
"electronic": {
"harm": "Contains toxic metals that contaminate soil and water.",
"handling": "Take to e-waste recycling centers or collection drives."
},
}


def classifyWaste(image_stream):
image=Image.open(image_stream).convert("RGB")
inputs=processor(images=image, return_tensors="pt")
with torch.no_grad():
outputs = model(**inputs)
predicted_class=outputs.logits.argmax(-1).item()
label = model.config.id2label[predicted_class].lower()
return label

def classifyClean(image_stream):
image=Image.open(image_stream).convert("RGB")
inputs=processor(images=image, return_tensors="pt")
with torch.no_grad():
outputs = model(**inputs)
logits = outputs.logits
probs = F.softmax(logits, dim=1)
confidence = probs.max().item()
return confidence

def extract_confidence_with_moviepy(video_stream):
# Save video to temporary file
with tempfile.NamedTemporaryFile(delete=False, suffix=".mp4") as temp_file:
temp_file.write(video_stream.read())
temp_path = temp_file.name

clip = VideoFileClip(temp_path)
duration = int(clip.duration) # in seconds
confidences = []

for t in range(0, duration):
frame = clip.get_frame(t) # returns a NumPy array
img = Image.fromarray(frame) # Convert to PIL Image
buffer = BytesIO()
img.save(buffer, format="JPEG")
buffer.seek(0)

result = classifyClean(buffer)
print(result)# Your function that returns confidence
confidences.append(result)

clip.close()
os.remove(temp_path)

if not confidences:
return 0.0

return sum(confidences) / len(confidences)


I tried using ai model for this classification but it is not working The code you've shared demonstrates a machine learning pipeline for classifying types of waste (such as organic, plastic, and electronic) using a pretrained transformer-based image classification model hosted on Hugging Face (watersplash/waste-classification). This code uses the transformers library by Hugging Face, PIL for image handling, and PyTorch (torch) for model inference. Here's a breakdown of its functionality in textual paragraph form:

Continue reading...