[HTML payload içeriği buraya]
27.4 C
Jakarta
Monday, November 25, 2024
HomeArtificial Intelligence
Artificial Intelligence

Naming and finding objects in photographs

By Admin
0
17


We’ve all change into used to deep studying’s success in picture classification. Higher Swiss Mountain canine or Bernese mountain canine? Pink panda or big panda? No downside.
Nevertheless, in actual life it’s not sufficient to call the one most salient object on an image. Prefer it or not, one of the crucial compelling examples is autonomous driving: We don’t need the algorithm to acknowledge simply that automotive in entrance of us, but in addition the pedestrian about to cross the road. And, simply detecting the pedestrian is just not ample. The precise location of objects issues.

The time period object detection is often used to discuss with the duty of naming and localizing a number of objects in a picture body. Object detection is troublesome; we’ll construct as much as it in a unfastened collection of posts, specializing in ideas as a substitute of aiming for final efficiency. Immediately, we’ll begin with a couple of simple constructing blocks: Classification, each single and a number of; localization; and mixing each classification and localization of a single object.

Dataset

We’ll be utilizing photographs and annotations from the Pascal VOC dataset which might be downloaded from this mirror.
Particularly, we’ll use knowledge from the 2007 problem and the identical JSON annotation file as used within the quick.ai course.

Fast obtain/group directions, shamelessly taken from a useful publish on the quick.ai wiki, are as follows:

# mkdir knowledge && cd knowledge
# curl -OL http://pjreddie.com/media/recordsdata/VOCtrainval_06-Nov-2007.tar
# curl -OL https://storage.googleapis.com/coco-dataset/exterior/PASCAL_VOC.zip
# tar -xf VOCtrainval_06-Nov-2007.tar
# unzip PASCAL_VOC.zip
# mv PASCAL_VOC/*.json .
# rmdir PASCAL_VOC
# tar -xvf VOCtrainval_06-Nov-2007.tar

In phrases, we take the pictures and the annotation file from completely different locations:

Whether or not you’re executing the listed instructions or arranging recordsdata manually, you must finally find yourself with directories/recordsdata analogous to those:

img_dir <- "knowledge/VOCdevkit/VOC2007/JPEGImages"
annot_file <- "knowledge/pascal_train2007.json"

Now we have to extract some info from that json file.

Preprocessing

Let’s shortly be certain that now we have all required libraries loaded.

Annotations include details about three forms of issues we’re fascinated with.

annotations <- fromJSON(file = annot_file)
str(annotations, max.degree = 1)
Listing of 4
 $ photographs     :Listing of 2501
 $ kind       : chr "situations"
 $ annotations:Listing of 7844
 $ classes :Listing of 20

First, traits of the picture itself (peak and width) and the place it’s saved. Not surprisingly, right here it’s one entry per picture.

Then, object class ids and bounding field coordinates. There could also be a number of of those per picture.
In Pascal VOC, there are 20 object lessons, from ubiquitous autos (automotive, aeroplane) over indispensable animals (cat, sheep) to extra uncommon (in well-liked datasets) sorts like potted plant or television monitor.

lessons <- c(
  "aeroplane",
  "bicycle",
  "hen",
  "boat",
  "bottle",
  "bus",
  "automotive",
  "cat",
  "chair",
  "cow",
  "diningtable",
  "canine",
  "horse",
  "bike",
  "particular person",
  "pottedplant",
  "sheep",
  "couch",
  "practice",
  "tvmonitor"
)

boxinfo <- annotations$annotations %>% {
  tibble(
    image_id = map_dbl(., "image_id"),
    category_id = map_dbl(., "category_id"),
    bbox = map(., "bbox")
  )
}

The bounding containers are actually saved in an inventory column and should be unpacked.

boxinfo <- boxinfo %>% 
  mutate(bbox = unlist(map(.$bbox, operate(x) paste(x, collapse = " "))))
boxinfo <- boxinfo %>% 
  separate(bbox, into = c("x_left", "y_top", "bbox_width", "bbox_height"))
boxinfo <- boxinfo %>% mutate_all(as.numeric)

For the bounding containers, the annotation file offers x_left and y_top coordinates, in addition to width and peak.
We’ll largely be working with nook coordinates, so we create the lacking x_right and y_bottom.

As common in picture processing, the y axis begins from the highest.

boxinfo <- boxinfo %>% 
  mutate(y_bottom = y_top + bbox_height - 1, x_right = x_left + bbox_width - 1)

Lastly, we nonetheless must match class ids to class names.

So, placing all of it collectively:

Word that right here nonetheless, now we have a number of entries per picture, every annotated object occupying its personal row.

There’s one step that may bitterly harm our localization efficiency if we later neglect it, so let’s do it now already: We have to scale all bounding field coordinates in response to the precise picture measurement we’ll use after we move it to our community.

target_height <- 224
target_width <- 224

imageinfo <- imageinfo %>% mutate(
  x_left_scaled = (x_left / image_width * target_width) %>% spherical(),
  x_right_scaled = (x_right / image_width * target_width) %>% spherical(),
  y_top_scaled = (y_top / image_height * target_height) %>% spherical(),
  y_bottom_scaled = (y_bottom / image_height * target_height) %>% spherical(),
  bbox_width_scaled =  (bbox_width / image_width * target_width) %>% spherical(),
  bbox_height_scaled = (bbox_height / image_height * target_height) %>% spherical()
)

Let’s take a look at our knowledge. Selecting one of many early entries and displaying the unique picture along with the article annotation yields

img_data <- imageinfo[4,]
img <- image_read(file.path(img_dir, img_data$file_name))
img <- image_draw(img)
rect(
  img_data$x_left,
  img_data$y_bottom,
  img_data$x_right,
  img_data$y_top,
  border = "white",
  lwd = 2
)
textual content(
  img_data$x_left,
  img_data$y_top,
  img_data$title,
  offset = 1,
  pos = 2,
  cex = 1.5,
  col = "white"
)
dev.off()

Now as indicated above, on this publish we’ll largely handle dealing with a single object in a picture. This implies now we have to determine, per picture, which object to single out.

An affordable technique appears to be selecting the article with the most important floor fact bounding field.

After this operation, we solely have 2501 photographs to work with – not many in any respect! For classification, we might merely use knowledge augmentation as supplied by Keras, however to work with localization we’d should spin our personal augmentation algorithm.
We’ll depart this to a later event and for now, deal with the fundamentals.

Lastly after train-test cut up

train_indices <- pattern(1:n_samples, 0.8 * n_samples)
train_data <- imageinfo_maxbb[train_indices,]
validation_data <- imageinfo_maxbb[-train_indices,]

our coaching set consists of 2000 photographs with one annotation every. We’re prepared to begin coaching, and we’ll begin gently, with single-object classification.

Single-object classification

In all instances, we are going to use XCeption as a fundamental function extractor. Having been skilled on ImageNet, we don’t count on a lot effective tuning to be essential to adapt to Pascal VOC, so we depart XCeption’s weights untouched

feature_extractor <-
  application_xception(
    include_top = FALSE,
    input_shape = c(224, 224, 3),
    pooling = "avg"
)

feature_extractor %>% freeze_weights()

and put just some customized layers on high.

mannequin <- keras_model_sequential() %>%
  feature_extractor %>%
  layer_batch_normalization() %>%
  layer_dropout(fee = 0.25) %>%
  layer_dense(items = 512, activation = "relu") %>%
  layer_batch_normalization() %>%
  layer_dropout(fee = 0.5) %>%
  layer_dense(items = 20, activation = "softmax")

mannequin %>% compile(
  optimizer = "adam",
  loss = "sparse_categorical_crossentropy",
  metrics = record("accuracy")
)

How ought to we move our knowledge to Keras? We might easy use Keras’ image_data_generator, however given we are going to want customized turbines quickly, we’ll construct a easy one ourselves.
This one delivers photographs in addition to the corresponding targets in a stream. Word how the targets usually are not one-hot-encoded, however integers – utilizing sparse_categorical_crossentropy as a loss operate allows this comfort.

batch_size <- 10

load_and_preprocess_image <- operate(image_name, target_height, target_width) {
  img_array <- image_load(
    file.path(img_dir, image_name),
    target_size = c(target_height, target_width)
    ) %>%
    image_to_array() %>%
    xception_preprocess_input() 
  dim(img_array) <- c(1, dim(img_array))
  img_array
}

classification_generator <-
  operate(knowledge,
           target_height,
           target_width,
           shuffle,
           batch_size) {
    i <- 1
    operate() {
      if (shuffle) {
        indices <- pattern(1:nrow(knowledge), measurement = batch_size)
      } else {
        if (i + batch_size >= nrow(knowledge))
          i <<- 1
        indices <- c(i:min(i + batch_size - 1, nrow(knowledge)))
        i <<- i + size(indices)
      }
      x <-
        array(0, dim = c(size(indices), target_height, target_width, 3))
      y <- array(0, dim = c(size(indices), 1))
      
      for (j in 1:size(indices)) {
        x[j, , , ] <-
          load_and_preprocess_image(knowledge[[indices[j], "file_name"]],
                                    target_height, target_width)
        y[j, ] <-
          knowledge[[indices[j], "category_id"]] - 1
      }
      x <- x / 255
      record(x, y)
    }
  }

train_gen <- classification_generator(
  train_data,
  target_height = target_height,
  target_width = target_width,
  shuffle = TRUE,
  batch_size = batch_size
)

valid_gen <- classification_generator(
  validation_data,
  target_height = target_height,
  target_width = target_width,
  shuffle = FALSE,
  batch_size = batch_size
)

Now how does coaching go?

mannequin %>% fit_generator(
  train_gen,
  epochs = 20,
  steps_per_epoch = nrow(train_data) / batch_size,
  validation_data = valid_gen,
  validation_steps = nrow(validation_data) / batch_size,
  callbacks = record(
    callback_model_checkpoint(
      file.path("class_only", "weights.{epoch:02d}-{val_loss:.2f}.hdf5")
    ),
    callback_early_stopping(persistence = 2)
  )
)

For us, after 8 epochs, accuracies on the practice resp. validation units have been at 0.68 and 0.74, respectively. Not too unhealthy given given we’re making an attempt to distinguish between 20 lessons right here.

Now let’s shortly suppose what we’d change if we have been to categorise a number of objects in a single picture. Adjustments largely concern preprocessing steps.

A number of object classification

This time, we multi-hot-encode our knowledge. For each picture (as represented by its filename), right here now we have a vector of size 20 the place 0 signifies absence, 1 means presence of the respective object class:

image_cats <- imageinfo %>% 
  choose(category_id) %>%
  mutate(category_id = category_id - 1) %>%
  pull() %>%
  to_categorical(num_classes = 20)

image_cats <- knowledge.body(image_cats) %>%
  add_column(file_name = imageinfo$file_name, .earlier than = TRUE)

image_cats <- image_cats %>% 
  group_by(file_name) %>% 
  summarise_all(.funs = funs(max))

n_samples <- nrow(image_cats)
train_indices <- pattern(1:n_samples, 0.8 * n_samples)
train_data <- image_cats[train_indices,]
validation_data <- image_cats[-train_indices,]

Correspondingly, we modify the generator to return a goal of dimensions batch_size * 20, as a substitute of batch_size * 1.

classification_generator <- 
  operate(knowledge,
           target_height,
           target_width,
           shuffle,
           batch_size) {
    i <- 1
    operate() {
      if (shuffle) {
        indices <- pattern(1:nrow(knowledge), measurement = batch_size)
      } else {
        if (i + batch_size >= nrow(knowledge))
          i <<- 1
        indices <- c(i:min(i + batch_size - 1, nrow(knowledge)))
        i <<- i + size(indices)
      }
      x <-
        array(0, dim = c(size(indices), target_height, target_width, 3))
      y <- array(0, dim = c(size(indices), 20))
      
      for (j in 1:size(indices)) {
        x[j, , , ] <-
          load_and_preprocess_image(knowledge[[indices[j], "file_name"]], 
                                    target_height, target_width)
        y[j, ] <-
          knowledge[indices[j], 2:21] %>% as.matrix()
      }
      x <- x / 255
      record(x, y)
    }
  }

train_gen <- classification_generator(
  train_data,
  target_height = target_height,
  target_width = target_width,
  shuffle = TRUE,
  batch_size = batch_size
)

valid_gen <- classification_generator(
  validation_data,
  target_height = target_height,
  target_width = target_width,
  shuffle = FALSE,
  batch_size = batch_size
)

Now, essentially the most attention-grabbing change is to the mannequin – although it’s a change to 2 traces solely.
Had been we to make use of categorical_crossentropy now (the non-sparse variant of the above), mixed with a softmax activation, we’d successfully inform the mannequin to select only one, specifically, essentially the most possible object.

As an alternative, we wish to determine: For every object class, is it current within the picture or not? Thus, as a substitute of softmax we use sigmoid, paired with binary_crossentropy, to acquire an unbiased verdict on each class.

feature_extractor <-
  application_xception(
    include_top = FALSE,
    input_shape = c(224, 224, 3),
    pooling = "avg"
  )

feature_extractor %>% freeze_weights()

mannequin <- keras_model_sequential() %>%
  feature_extractor %>%
  layer_batch_normalization() %>%
  layer_dropout(fee = 0.25) %>%
  layer_dense(items = 512, activation = "relu") %>%
  layer_batch_normalization() %>%
  layer_dropout(fee = 0.5) %>%
  layer_dense(items = 20, activation = "sigmoid")

mannequin %>% compile(optimizer = "adam",
                  loss = "binary_crossentropy",
                  metrics = record("accuracy"))

And at last, once more, we match the mannequin:

mannequin %>% fit_generator(
  train_gen,
  epochs = 20,
  steps_per_epoch = nrow(train_data) / batch_size,
  validation_data = valid_gen,
  validation_steps = nrow(validation_data) / batch_size,
  callbacks = record(
    callback_model_checkpoint(
      file.path("multiclass", "weights.{epoch:02d}-{val_loss:.2f}.hdf5")
    ),
    callback_early_stopping(persistence = 2)
  )
)

This time, (binary) accuracy surpasses 0.95 after one epoch already, on each the practice and validation units. Not surprisingly, accuracy is considerably increased right here than after we needed to single out considered one of 20 lessons (and that, with different confounding objects current generally!).

Now, chances are high that should you’ve completed any deep studying earlier than, you’ve completed picture classification in some type, maybe even within the multiple-object variant. To construct up within the course of object detection, it’s time we add a brand new ingredient: localization.

Single-object localization

From right here on, we’re again to coping with a single object per picture. So the query now could be, how can we be taught bounding containers?
For those who’ve by no means heard of this, the reply will sound unbelievably easy (naive even): We formulate this as a regression downside and goal to foretell the precise coordinates. To set lifelike expectations – we certainly shouldn’t count on final precision right here. However in a manner it’s wonderful it does even work in any respect.

What does this imply, formulate as a regression downside? Concretely, it means we’ll have a dense output layer with 4 items, every equivalent to a nook coordinate.

So let’s begin with the mannequin this time. Once more, we use Xception, however there’s an essential distinction right here: Whereas earlier than, we mentioned pooling = "avg" to acquire an output tensor of dimensions batch_size * variety of filters, right here we don’t do any averaging or flattening out of the spatial grid. It’s because it’s precisely the spatial info we’re fascinated with!

For Xception, the output decision might be 7×7. So a priori, we shouldn’t count on excessive precision on objects a lot smaller than about 32×32 pixels (assuming the usual enter measurement of 224×224).

feature_extractor <- application_xception(
  include_top = FALSE,
  input_shape = c(224, 224, 3)
)

feature_extractor %>% freeze_weights()

Now we append our customized regression module.

mannequin <- keras_model_sequential() %>%
  feature_extractor %>%
  layer_flatten() %>%
  layer_batch_normalization() %>%
  layer_dropout(fee = 0.25) %>%
  layer_dense(items = 512, activation = "relu") %>%
  layer_batch_normalization() %>%
  layer_dropout(fee = 0.5) %>%
  layer_dense(items = 4)

We’ll practice with one of many loss capabilities widespread in regression duties, imply absolute error. However in duties like object detection or segmentation, we’re additionally fascinated with a extra tangible amount: How a lot do estimate and floor fact overlap?

Overlap is normally measured as Intersection over Union, or Jaccard distance. Intersection over Union is precisely what it says, a ratio between house shared by the objects and house occupied after we take them collectively.

To evaluate the mannequin’s progress, we will simply code this as a customized metric:

metric_iou <- operate(y_true, y_pred) {
  
  # order is [x_left, y_top, x_right, y_bottom]
  intersection_xmin <- k_maximum(y_true[ ,1], y_pred[ ,1])
  intersection_ymin <- k_maximum(y_true[ ,2], y_pred[ ,2])
  intersection_xmax <- k_minimum(y_true[ ,3], y_pred[ ,3])
  intersection_ymax <- k_minimum(y_true[ ,4], y_pred[ ,4])
  
  area_intersection <- (intersection_xmax - intersection_xmin) * 
                       (intersection_ymax - intersection_ymin)
  area_y <- (y_true[ ,3] - y_true[ ,1]) * (y_true[ ,4] - y_true[ ,2])
  area_yhat <- (y_pred[ ,3] - y_pred[ ,1]) * (y_pred[ ,4] - y_pred[ ,2])
  area_union <- area_y + area_yhat - area_intersection
  
  iou <- area_intersection/area_union
  k_mean(iou)
  
}

Mannequin compilation then goes like

mannequin %>% compile(
  optimizer = "adam",
  loss = "mae",
  metrics = record(custom_metric("iou", metric_iou))
)

Now modify the generator to return bounding field coordinates as targets…

localization_generator <-
  operate(knowledge,
           target_height,
           target_width,
           shuffle,
           batch_size) {
    i <- 1
    operate() {
      if (shuffle) {
        indices <- pattern(1:nrow(knowledge), measurement = batch_size)
      } else {
        if (i + batch_size >= nrow(knowledge))
          i <<- 1
        indices <- c(i:min(i + batch_size - 1, nrow(knowledge)))
        i <<- i + size(indices)
      }
      x <-
        array(0, dim = c(size(indices), target_height, target_width, 3))
      y <- array(0, dim = c(size(indices), 4))
      
      for (j in 1:size(indices)) {
        x[j, , , ] <-
          load_and_preprocess_image(knowledge[[indices[j], "file_name"]], 
                                    target_height, target_width)
        y[j, ] <-
          knowledge[indices[j], c("x_left_scaled",
                             "y_top_scaled",
                             "x_right_scaled",
                             "y_bottom_scaled")] %>% as.matrix()
      }
      x <- x / 255
      record(x, y)
    }
  }

train_gen <- localization_generator(
  train_data,
  target_height = target_height,
  target_width = target_width,
  shuffle = TRUE,
  batch_size = batch_size
)

valid_gen <- localization_generator(
  validation_data,
  target_height = target_height,
  target_width = target_width,
  shuffle = FALSE,
  batch_size = batch_size
)

… and we’re able to go!

mannequin %>% fit_generator(
  train_gen,
  epochs = 20,
  steps_per_epoch = nrow(train_data) / batch_size,
  validation_data = valid_gen,
  validation_steps = nrow(validation_data) / batch_size,
  callbacks = record(
    callback_model_checkpoint(
      file.path("loc_only", "weights.{epoch:02d}-{val_loss:.2f}.hdf5")
    ),
    callback_early_stopping(persistence = 2)
  )
)

After 8 epochs, IOU on each coaching and take a look at units is round 0.35. This quantity doesn’t look too good. To be taught extra about how coaching went, we have to see some predictions. Right here’s a comfort operate that shows a picture, the bottom fact field of essentially the most salient object (as outlined above), and if given, class and bounding field predictions.

plot_image_with_boxes <- operate(file_name,
                                  object_class,
                                  field,
                                  scaled = FALSE,
                                  class_pred = NULL,
                                  box_pred = NULL) {
  img <- image_read(file.path(img_dir, file_name))
  if(scaled) img <- image_resize(img, geometry = "224x224!")
  img <- image_draw(img)
  x_left <- field[1]
  y_bottom <- field[2]
  x_right <- field[3]
  y_top <- field[4]
  rect(
    x_left,
    y_bottom,
    x_right,
    y_top,
    border = "cyan",
    lwd = 2.5
  )
  textual content(
    x_left,
    y_top,
    object_class,
    offset = 1,
    pos = 2,
    cex = 1.5,
    col = "cyan"
  )
  if (!is.null(box_pred))
    rect(box_pred[1],
         box_pred[2],
         box_pred[3],
         box_pred[4],
         border = "yellow",
         lwd = 2.5)
  if (!is.null(class_pred))
    textual content(
      box_pred[1],
      box_pred[2],
      class_pred,
      offset = 0,
      pos = 4,
      cex = 1.5,
      col = "yellow")
  dev.off()
  img %>% image_write(paste0("preds_", file_name))
  plot(img)
}

First, let’s see predictions on pattern photographs from the coaching set.

train_1_8 <- train_data[1:8, c("file_name",
                               "name",
                               "x_left_scaled",
                               "y_top_scaled",
                               "x_right_scaled",
                               "y_bottom_scaled")]

for (i in 1:8) {
  preds <-
    mannequin %>% predict(
      load_and_preprocess_image(train_1_8[i, "file_name"], 
                                target_height, target_width),
      batch_size = 1
  )
  plot_image_with_boxes(train_1_8$file_name[i],
                        train_1_8$title[i],
                        train_1_8[i, 3:6] %>% as.matrix(),
                        scaled = TRUE,
                        box_pred = preds)
}
Sample bounding box predictions on the training set.

As you’d guess from trying, the cyan-colored containers are the bottom fact ones. Now trying on the predictions explains rather a lot concerning the mediocre IOU values! Let’s take the very first pattern picture – we needed the mannequin to deal with the couch, however it picked the desk, which can be a class within the dataset (though within the type of eating desk). Related with the picture on the appropriate of the primary row – we needed to it to select simply the canine however it included the particular person, too (by far essentially the most regularly seen class within the dataset).
So we truly made the duty much more troublesome than had we stayed with e.g., ImageNet the place usually a single object is salient.

Now test predictions on the validation set.

Some bounding box predictions on the validation set.

Once more, we get the same impression: The mannequin did be taught one thing, however the process is sick outlined. Have a look at the third picture in row 2: Isn’t it fairly consequent the mannequin picks all folks as a substitute of singling out some particular man?

If single-object localization is that easy, how technically concerned can it’s to output a category label on the identical time?
So long as we stick with a single object, the reply certainly is: not a lot.

Let’s end up immediately with a constrained mixture of classification and localization: detection of a single object.

Single-object detection

Combining regression and classification into one means we’ll wish to have two outputs in our mannequin.
We’ll thus use the practical API this time.
In any other case, there isn’t a lot new right here: We begin with an XCeption output of spatial decision 7×7, append some customized processing and return two outputs, one for bounding field regression and one for classification.

feature_extractor <- application_xception(
  include_top = FALSE,
  input_shape = c(224, 224, 3)
)

enter <- feature_extractor$enter
widespread <- feature_extractor$output %>%
  layer_flatten(title = "flatten") %>%
  layer_activation_relu() %>%
  layer_dropout(fee = 0.25) %>%
  layer_dense(items = 512, activation = "relu") %>%
  layer_batch_normalization() %>%
  layer_dropout(fee = 0.5)

regression_output <-
  layer_dense(widespread, items = 4, title = "regression_output")
class_output <- layer_dense(
  widespread,
  items = 20,
  activation = "softmax",
  title = "class_output"
)

mannequin <- keras_model(
  inputs = enter,
  outputs = record(regression_output, class_output)
)

When defining the losses (imply absolute error and categorical crossentropy, simply as within the respective single duties of regression and classification), we might weight them so that they find yourself on roughly a typical scale. In reality that didn’t make a lot of a distinction so we present the respective code in commented type.

mannequin %>% freeze_weights(to = "flatten")

mannequin %>% compile(
  optimizer = "adam",
  loss = record("mae", "sparse_categorical_crossentropy"),
  #loss_weights = record(
  #  regression_output = 0.05,
  #  class_output = 0.95),
  metrics = record(
    regression_output = custom_metric("iou", metric_iou),
    class_output = "accuracy"
  )
)

Identical to mannequin outputs and losses are each lists, the information generator has to return the bottom fact samples in an inventory.
Becoming the mannequin then goes as common.

Previous articleSaaS Safety Posture—It’s not you, it’s me!
Next articleGreatest Purchase’s Weekend Sale Takes $130 Off Almost Each ninth Gen iPad, Beginning at Simply $199.99
Adminhttps://dwipanks.xyz

Related Articles

LEAVE A REPLY

Please enter your comment!
Please enter your name here

Latest Articles

Load more

ABOUT US

Dwipanks is your Technology, Nanotechnology, Telecom, Cloud Computing, Artificial Intelligence, Mobile, Robotics, Drone, Big Data and 3D Printing related news website. We provide you with the latest breaking news and videos straight from the Tech industry. For more latest updates keep visiting us, Thankyou.

Copyright © 2024 - dwipanks.xyz. All rights reserved.