app.py

import json
import os
import random
from functools import partial
from pathlib import Path
from typing import List

import deepinv as dinv
import gradio as gr
import torch
from PIL import Image
from torchvision import transforms

from factories import PhysicsWithGenerator, EvalModel, BaselineModel, EvalDataset, Metric


### Config
DEVICE_STR = 'cuda'            # run model inference on NVIDIA gpu
torch.set_grad_enabled(False)  # stops tracking values for gradients


### Gradio Utils

def generate_imgs_from_user(image,
                            model: EvalModel, baseline: BaselineModel,
                            physics: PhysicsWithGenerator, use_gen: bool,
                            metrics: List[Metric]):
    if image is None:
        return None, None, None, None, None, None, None, None

    # PIL image -> torch.Tensor
    x = transforms.ToTensor()(image).unsqueeze(0).to(DEVICE_STR)

    return generate_imgs(x, model, baseline, physics, use_gen, metrics)

def generate_imgs_from_dataset(dataset: EvalDataset, idx: int,
                               model: EvalModel, baseline: BaselineModel,
                               physics: PhysicsWithGenerator, use_gen: bool,
                               metrics: List[Metric]):
    ### Load 1 image
    x = dataset[idx]    # shape : (C, H, W)
    x = x.unsqueeze(0)  # shape : (1, C, H, W)

    return generate_imgs(x, model, baseline, physics, use_gen, metrics)

def generate_random_imgs_from_dataset(dataset: EvalDataset,
                                      model: EvalModel,
                                      baseline: BaselineModel,
                                      physics: PhysicsWithGenerator,
                                      use_gen: bool,
                                      metrics: List[Metric]):
    idx = random.randint(0, len(dataset)-1)
    x, y, out, out_baseline, saved_params_str, metrics_y, metrics_out, metrics_out_baseline = generate_imgs_from_dataset(
        dataset, idx, model, baseline, physics, use_gen, metrics
        )
    return idx, x, y, out, out_baseline, saved_params_str, metrics_y, metrics_out, metrics_out_baseline
    
def generate_imgs(x: torch.Tensor,
                  model: EvalModel, baseline: BaselineModel,
                  physics: PhysicsWithGenerator, use_gen: bool,
                  metrics: List[Metric]):

    with torch.no_grad():
        ### Compute y
        y = physics(x, use_gen)  # possible reduction in img shape due to Blurring

        ### Compute x_hat
        out = model(y=y, physics=physics.physics)
        out_baseline = baseline(y=y, physics=physics.physics)

        ### Process tensors before metric computation
        if "Blur" in physics.name:
            w_1, w_2 = (x.shape[2] - y.shape[2]) // 2, (x.shape[2] + y.shape[2]) // 2
            h_1, h_2 = (x.shape[3] - y.shape[3]) // 2, (x.shape[3] + y.shape[3]) // 2

            x = x[..., w_1:w_2, h_1:h_2]
            out = out[..., w_1:w_2, h_1:h_2]
            if out_baseline.shape != out.shape:
                out_baseline = out_baseline[..., w_1:w_2, h_1:h_2]

        ### Metrics
        metrics_y = ""
        metrics_out = ""
        metrics_out_baseline = ""
        for metric in metrics:
            if y.shape == x.shape:
                metrics_y += f"{metric.name} = {metric(y, x).item():.4f}" + "\n"
            metrics_out += f"{metric.name} = {metric(out, x).item():.4f}" + "\n"
            metrics_out_baseline += f"{metric.name} = {metric(out_baseline, x).item():.4f}" + "\n"

    ### Process y when y shape is different from x shape
    if physics.name == "MRI" in physics.name:
        y_plot = physics.physics.prox_l2(physics.physics.A_adjoint(y), y, 1e4)
    else:
        y_plot = y.clone()

    ### Processing images for plotting :
    #     - clip value outside of [0,1]
    #     - shape (1, C, H, W) -> (C, H, W)
    #     - torch.Tensor object -> Pil object
    process_img = partial(dinv.utils.plotting.preprocess_img, rescale_mode="clip")
    to_pil = transforms.ToPILImage()
    x = to_pil(process_img(x)[0].to('cpu'))
    y = to_pil(process_img(y_plot)[0].to('cpu'))
    out = to_pil(process_img(out)[0].to('cpu'))
    out_baseline = to_pil(process_img(out_baseline)[0].to('cpu'))

    return x, y, out, out_baseline, physics.display_saved_params(), metrics_y, metrics_out, metrics_out_baseline


get_list_metrics_on_DEVICE_STR = partial(Metric.get_list_metrics, device_str=DEVICE_STR)
get_eval_model_on_DEVICE_STR = partial(EvalModel, device_str=DEVICE_STR)
get_baseline_model_on_DEVICE_STR = partial(BaselineModel, device_str=DEVICE_STR)
get_dataset_on_DEVICE_STR = partial(EvalDataset, device_str=DEVICE_STR)
get_physics_on_DEVICE_STR = partial(PhysicsWithGenerator, device_str=DEVICE_STR)

AVAILABLE_PHYSICS = ['MotionBlur_easy', 'MotionBlur_medium', 'MotionBlur_hard',
                     'GaussianBlur_easy', 'GaussianBlur_medium', 'GaussianBlur_hard']
def get_dataset(dataset_name):
    global AVAILABLE_PHYSICS
    if dataset_name == 'MRI':
        AVAILABLE_PHYSICS = ['MRI']
        baseline_name = 'DPIR_MRI'
        physics_name = 'MRI'
    elif dataset_name == 'CT':
        AVAILABLE_PHYSICS = ['CT']
        baseline_name = 'DPIR_CT'
        physics_name = 'CT'
    else:
        AVAILABLE_PHYSICS = ['MotionBlur_easy', 'MotionBlur_medium', 'MotionBlur_hard',
                             'GaussianBlur_easy', 'GaussianBlur_medium', 'GaussianBlur_hard']
        baseline_name = 'DPIR'
        physics_name = 'MotionBlur_easy'

    dataset = get_dataset_on_DEVICE_STR(dataset_name)
    physics = get_physics_on_DEVICE_STR(physics_name)
    baseline = get_baseline_model_on_DEVICE_STR(baseline_name)
    return dataset, physics, baseline


### Gradio Blocks interface

# Define custom CSS
custom_css = """
.fixed-textbox textarea {
    height: 100px !important;  /* Adjust height to fit exactly 4 lines */
    overflow: scroll;          /* Add a scroll bar if necessary */
    resize: none;              /* User can resize vertically the textbox */
}
"""

title = "Inverse problem playground"  # displayed on gradio tab and in the gradio page
with gr.Blocks(title=title, css=custom_css) as interface:
    gr.Markdown("## " + title)

    # DEFAULT VALUES
    # Issue: giving directly a `torch.nn.module` to `gr.State(...)` since it has __call__ method
    # Solution: using lambda expression
    model_a_placeholder = gr.State(lambda: get_eval_model_on_DEVICE_STR("unext_emb_physics_config_C", ""))
    model_b_placeholder = gr.State(lambda: get_baseline_model_on_DEVICE_STR("DPIR"))
    dataset_placeholder = gr.State(get_dataset_on_DEVICE_STR("Natural"))
    physics_placeholder = gr.State(lambda: get_physics_on_DEVICE_STR("MotionBlur_easy"))
    idx_placeholder = gr.State(0)

    metric_names = ["PSNR"]
    metrics_placeholder = gr.State(get_list_metrics_on_DEVICE_STR(metric_names))

    @gr.render(inputs=[dataset_placeholder, physics_placeholder])
    def dynamic_layout(dataset, physics):
        ### LAYOUT

        # Display images
        with gr.Row():
            gt_img = gr.Image(label=f"Ground-truth IMAGE", interactive=True)
            observed_img = gr.Image(label=f"Observed IMAGE", interactive=False)
            model_a_out = gr.Image(label="RAM OUTPUT", interactive=False)
            model_b_out = gr.Image(label="DPIR OUTPUT", interactive=False)

        # Manage datasets and display metric values
        with gr.Row():
            with gr.Column():
                run_button = gr.Button("Demo on above image")
                choose_dataset = gr.Radio(choices=EvalDataset.all_datasets,
                                          label="Datasets",
                                          value=dataset.name)
                idx_slider = gr.Slider(minimum=0, maximum=len(dataset)-1, step=1, label="Sample index")
                with gr.Row():
                    load_button = gr.Button("Run on index image from dataset")
                    load_random_button = gr.Button("Run on random image from dataset")
            with gr.Column():
                observed_metrics = gr.Textbox(label="PSNR(Observed, Ground-truth)",
                                              elem_classes=["fixed-textbox"])
            with gr.Column():
                out_a_metric = gr.Textbox(label="PSNR(RAM(Observed, Ground-truth)",
                                          elem_classes=["fixed-textbox"])
            with gr.Column():
                out_b_metric = gr.Textbox(label="PSNR(DPIR(Observed, Ground-truth)",
                                          elem_classes=["fixed-textbox"])

        # Manage physics
        with gr.Row():
            with gr.Column(scale=1):
                choose_physics = gr.Radio(choices=AVAILABLE_PHYSICS,
                                          label="Physics",
                                          value=physics.name)
                use_generator_button = gr.Checkbox(label="Generate physics parameters during inference")
            with gr.Column(scale=1):
                with gr.Row():
                    key_selector = gr.Dropdown(choices=list(physics.saved_params["updatable_params"].keys()),
                                                label="Updatable Parameter Key")
                    value_text = gr.Textbox(label="Update Value")
                update_button = gr.Button("Manually update parameter value")
            with gr.Column(scale=2):
                physics_params = gr.Textbox(label="Physics parameters",
                                            elem_classes=["fixed-textbox"],
                                            value=physics.display_saved_params())  


        ### Event listeners

        choose_dataset.change(fn=get_dataset,
                              inputs=choose_dataset,
                              outputs=[dataset_placeholder, physics_placeholder, model_b_placeholder])
        choose_physics.change(fn=get_physics_on_DEVICE_STR,
                              inputs=choose_physics,
                              outputs=[physics_placeholder])
        update_button.click(fn=physics.update_and_display_params,
                            inputs=[key_selector, value_text], outputs=physics_params)
        run_button.click(fn=generate_imgs_from_user,
                         inputs=[gt_img,
                                 model_a_placeholder,
                                 model_b_placeholder,
                                 physics_placeholder,
                                 use_generator_button,
                                 metrics_placeholder],
                         outputs=[gt_img, observed_img, model_a_out, model_b_out,
                                  physics_params, observed_metrics, out_a_metric, out_b_metric])
        load_button.click(fn=generate_imgs_from_dataset,
                          inputs=[dataset_placeholder,
                                  idx_slider,
                                  model_a_placeholder,
                                  model_b_placeholder,
                                  physics_placeholder,
                                  use_generator_button,
                                  metrics_placeholder],
                          outputs=[gt_img, observed_img, model_a_out, model_b_out,
                                   physics_params, observed_metrics, out_a_metric, out_b_metric])
        load_random_button.click(fn=generate_random_imgs_from_dataset,
                                 inputs=[dataset_placeholder,
                                         model_a_placeholder,
                                         model_b_placeholder,
                                         physics_placeholder,
                                         use_generator_button,
                                         metrics_placeholder],
                                 outputs=[idx_slider, gt_img, observed_img, model_a_out, model_b_out,
                                          physics_params, observed_metrics, out_a_metric, out_b_metric])


interface.launch()