app.py

from pathlib import Path
import gradio as gr
import torch
from transformers import AutoModelForImageClassification
import shutil
from optimum.pipelines import pipeline
import os, cv2, numpy as np
from PIL import Image
import itertools
device = 1 if torch.cuda.is_available() else "cpu"

chk_point = "kdhht2334/autotrain-diffusion-emotion-facial-expression-recognition-40429105176"

model = AutoModelForImageClassification.from_pretrained(chk_point)

##Add face detector
protoPath = os.path.sep.join(["face_detector", "deploy.prototxt"])
modelPath = os.path.sep.join(["face_detector",
                                      "res10_300x300_ssd_iter_140000.caffemodel"])
net = cv2.dnn.readNetFromCaffe(protoPath, modelPath)


def draw_results_ssd(detected, input_img, ad):
    img_w, img_h, _ = np.shape(input_img)
    # loop over the detections
    cropped_output = []
    if detected.shape[2] > 0:
        print("detected face: {}".format(detected.shape[2]))
        for i in range(0, detected.shape[2]):
            # extract the confidence (i.e., probability) associated with the
            # prediction
            confidence = detected[0, 0, i, 2]

            # filter out weak detections
            if confidence > 0.5:
                # compute the (x, y)-coordinates of the bounding box for
                # the face and extract the face ROI
                (h0, w0) = input_img.shape[:2]
                box = detected[0, 0, i, 3:7] * np.array([w0, h0, w0, h0])
                (startX, startY, endX, endY) = box.astype("int")
                # print((startX, startY, endX, endY))
                x1 = startX
                y1 = startY
                w = endX - startX
                h = endY - startY

                x2 = x1 + w
                y2 = y1 + h

                xw1 = max(int(x1 - ad * w), 0)
                yw1 = max(int(y1 - ad * h), 0)
                xw2 = min(int(x2 + ad * w), img_w - 1)
                yw2 = min(int(y2 + ad * h), img_h - 1)

                face_crop = input_img[yw1:yw2, xw1:xw2]
                if face_crop.size != 0:
                    cropped_output.append(face_crop)
    return cropped_output


emotion_dict = {
'neutral': '0',
'happy': '1',
'sad' :'2',
'surprise': '3',
'fear': '4',
'disgust': '5',
'angry': '6',
'uncertain': '7',
'nonface': '8',
}


try:
    pipe = pipeline(
        "image-classification",
        chk_point,
        accelerator="bettertransformer",
        device=device,
    )
except NotImplementedError:
    from transformers import pipeline

    pipe = pipeline("image-classification", chk_point, device=device)


# def make_label_folders():
#     folders = model.config.label2id.keys()
#     for folder in folders:
#         folder = Path(folder)
#         if not folder.exists():
#             folder.mkdir()
#     return folders


# def predictions_into_folders(files):
#     files = [file.name for file in files]
#     files = [
#         file for file in files if not file.startswith(".") and "DS_Store" not in file
#     ]
#     folders = make_label_folders()
#     predictions = pipe(files)
#     for file, prediction in zip(files, predictions):
#         label = prediction[0]["label"]
#         file_name = Path(file).name
#         shutil.copy(file, f"{label}/{file_name}")
#     for folder in folders:
#         shutil.make_archive(folder, "zip", ".", folder)
#     return [f"{folder}.zip" for folder in folders]

# demo = gr.Interface(
#     predictions_into_folders,
#     gr.Files(file_count="directory", file_types=["image"]),
#     gr.Files(),
#     cache_examples=True,
# )
# demo.launch(enable_queue=True)

def face_detector(input_img):
    img_h, img_w, _ = np.shape(input_img)
    blob = cv2.dnn.blobFromImage(cv2.resize(input_img, (300, 300)), 1.0,
                                 (300, 300), (104.0, 177.0, 123.0))
    net.setInput(blob)
    detected = net.forward()  # For example, (1, 1, 200, 7)
    cropped_list = draw_results_ssd(detected, input_img, 0.1 )  # 128
    return cropped_list

def predict(image):
    print(type(image))
    cropped_list = face_detector(image)
    print(cropped_list)
    output = [None for _ in range(20)]
    for i in range(len(cropped_list)):
        predictions = pipe(Image.fromarray(cropped_list[i]))
        output[int(i*2)] = cropped_list[i]
        output[int(i*2+1)] = {p["label"]: p["score"] for p in predictions}
    return output

gr.Interface(
    predict,
    inputs=gr.inputs.Image(label="Upload image"),
    outputs=list(itertools.chain(*[[gr.outputs.Image(label="image {}".format(ind)), gr.outputs.Label(num_top_classes=7, label="Result {}".format(ind))] for ind in range(10)])),
    examples=[["examples/9_peoples.jpg"], ["examples/sad.jpg"], ["examples/angry.jpg"], ["examples/surprise.jpg"]],
    title="FER trained on DiffusionFER dataset",
).launch()