app.py

# import gradio as gr
# #import torch
# import yolov7


# #
# # from huggingface_hub import hf_hub_download
# from huggingface_hub import HfApi


# # Images
# #torch.hub.download_url_to_file('https://github.com/ultralytics/yolov5/raw/master/data/images/zidane.jpg', 'zidane.jpg')
# #torch.hub.download_url_to_file('https://raw.githubusercontent.com/obss/sahi/main/tests/data/small-vehicles1.jpeg', 'small-vehicles1.jpeg')
    
# def yolov7_inference(
#     image: gr.inputs.Image = None,
#     model_path: gr.inputs.Dropdown = None,
#     image_size: gr.inputs.Slider = 640,
#     conf_threshold: gr.inputs.Slider = 0.25,
#     iou_threshold: gr.inputs.Slider = 0.45,
# ):
#     """
#     YOLOv7 inference function
#     Args:
#         image: Input image
#         model_path: Path to the model
#         image_size: Image size
#         conf_threshold: Confidence threshold
#         iou_threshold: IOU threshold
#     Returns:
#         Rendered image
#     """

#     model = yolov7.load(model_path, device="cpu", hf_model=True, trace=False)
#     model.conf = conf_threshold
#     model.iou = iou_threshold
#     results = model([image], size=image_size)
#     return results.render()[0]
        


# inputs = [
#     gr.inputs.Image(type="pil", label="Input Image"),
#     gr.inputs.Dropdown(
#         choices=[
#             "alshimaa/model_baseline",
#             "alshimaa/model_yolo7",
#             #"kadirnar/yolov7-v0.1",
#         ],
#         default="alshimaa/model_baseline",
#         label="Model",
#     )
#     #gr.inputs.Slider(minimum=320, maximum=1280, default=640, step=32, label="Image Size")
#     #gr.inputs.Slider(minimum=0.0, maximum=1.0, default=0.25, step=0.05, label="Confidence Threshold"),
#     #gr.inputs.Slider(minimum=0.0, maximum=1.0, default=0.45, step=0.05, label="IOU Threshold")
# ]

# outputs = gr.outputs.Image(type="filepath", label="Output Image")
# title = "Smart Environmental Eye (SEE)"

# examples = [['image1.jpg', 'alshimaa/model_yolo7', 640, 0.25, 0.45], ['image2.jpg', 'alshimaa/model_yolo7', 640, 0.25, 0.45], ['image3.jpg', 'alshimaa/model_yolo7', 640, 0.25, 0.45]]
# demo_app = gr.Interface(
#     fn=yolov7_inference,
#     inputs=inputs,
#     outputs=outputs,
#     title=title,
#     examples=examples,
#     cache_examples=True,
#     theme='huggingface',
# )

# demo_app.launch(debug=True, enable_queue=True)



import subprocess
import tempfile
import time
from pathlib import Path

import cv2
import gradio as gr

from inferer import Inferer

pipeline = Inferer("alshimaa/model_baseline", device='cuda')


def fn_image(image, conf_thres, iou_thres):
    return pipeline(image, conf_thres, iou_thres)


def fn_video(video_file, conf_thres, iou_thres, start_sec, duration):
    start_timestamp = time.strftime("%H:%M:%S", time.gmtime(start_sec))
    end_timestamp = time.strftime("%H:%M:%S", time.gmtime(start_sec + duration))

    suffix = Path(video_file).suffix

    clip_temp_file = tempfile.NamedTemporaryFile(suffix=suffix)
    subprocess.call(
        f"ffmpeg -y -ss {start_timestamp} -i {video_file} -to {end_timestamp} -c copy {clip_temp_file.name}".split()
    )

    # Reader of clip file
    cap = cv2.VideoCapture(clip_temp_file.name)

    # This is an intermediary temp file where we'll write the video to
    # Unfortunately, gradio doesn't play too nice with videos rn so we have to do some hackiness
    # with ffmpeg at the end of the function here.
    with tempfile.NamedTemporaryFile(suffix=".mp4") as temp_file:
        out = cv2.VideoWriter(temp_file.name, cv2.VideoWriter_fourcc(*"MP4V"), 30, (1280, 720))

        num_frames = 0
        max_frames = duration * 30
        while cap.isOpened():
            try:
                ret, frame = cap.read()
                if not ret:
                    break
            except Exception as e:
                print(e)
                continue
            print("FRAME DTYPE", type(frame))
            out.write(pipeline(frame, conf_thres, iou_thres))
            num_frames += 1
            print("Processed {} frames".format(num_frames))
            if num_frames == max_frames:
                break

        out.release()

        # Aforementioned hackiness
        out_file = tempfile.NamedTemporaryFile(suffix="out.mp4", delete=False)
        subprocess.run(f"ffmpeg -y -loglevel quiet -stats -i {temp_file.name} -c:v libx264 {out_file.name}".split())

    return out_file.name


image_interface = gr.Interface(
    fn=fn_image,
    inputs=[
        "image",
        gr.Slider(0, 1, value=0.5, label="Confidence Threshold"),
        gr.Slider(0, 1, value=0.5, label="IOU Threshold"),
    ],
    outputs=gr.Image(type="file"),
    examples=[["image1.jpg", 0.5, 0.5], ["image2.jpg", 0.25, 0.45], ["image3.jpg", 0.25, 0.45]],
    title="Smart Environmental Eye (SEE)",
    allow_flagging=False,
    allow_screenshot=False,
)

video_interface = gr.Interface(
    fn=fn_video,
    inputs=[
        gr.Video(type="file"),
        gr.Slider(0, 1, value=0.25, label="Confidence Threshold"),
        gr.Slider(0, 1, value=0.45, label="IOU Threshold"),
        gr.Slider(0, 10, value=0, label="Start Second", step=1),
        gr.Slider(0, 10 if pipeline.device.type != 'cpu' else 3, value=4, label="Duration", step=1),
    ],
    outputs=gr.Video(type="file", format="mp4"),
    # examples=[
    #     ["video.mp4", 0.25, 0.45, 0, 2],
       
    # ],
    title="Smart Environmental Eye (SEE)",
    allow_flagging=False,
    allow_screenshot=False,
)



if __name__ == "__main__":
    gr.TabbedInterface(
        [image_interface, video_interface],
        ["Run on Images", "Run on Videos"],
    ).launch()