I tried to detect an object with M2Det!

things to do

-Previous time and Last time performed object detection with SSD and YOLO v3, respectively. I tried, but this time I will try to detect the object with M2Det --This time, we will run it on Google Colaboratory.

Overview

1. Preparation of execution environment (executed by Google Colaboratory)
2. Download the model from Google Drive
3. Copy image file
4. Run the model
5. Display results

Execution environment

• google colaboratory
• pytorch
• opencv
• tqdm
• addict

1. Preparation of execution environment (executed by Google Colaboratory)

--Open Google Colaboratory and change "Runtime Type Change" to "GPU" on the "Runtime" tab.

――After that, do the following --Install various packages --Clone repository --Run shell

Preparation of execution environment

!pip install torch torchvision
!pip install opencv-python tqdm addict
!git clone https://github.com/qijiezhao/M2Det.git
%cd M2Det/
!sh make.sh

2. Download the model from Google Drive

-The README of GitHub has the link destination of the trained model (https://drive.google.com/file/d/1NM1UDdZnwHwiNDxhcP-nndaWj24m- 90L / view), so I will download it on the code from this Google Drive link --For download, I referred to the code of here. --You can download by command by executing download_file_from_google_drive.

python

import requests

def download_file_from_google_drive(id, destination):
    URL = "https://docs.google.com/uc?export=download"

    session = requests.Session()

    response = session.get(URL, params = { 'id' : id }, stream = True)
    token = get_confirm_token(response)

    if token:
        params = { 'id' : id, 'confirm' : token }
        response = session.get(URL, params = params, stream = True)

    save_response_content(response, destination)    

def get_confirm_token(response):
    for key, value in response.cookies.items():
        if key.startswith('download_warning'):
            return value

    return None

def save_response_content(response, destination):
    CHUNK_SIZE = 32768

    with open(destination, "wb") as f:
        for chunk in response.iter_content(CHUNK_SIZE):
            if chunk: # filter out keep-alive new chunks
                f.write(chunk)

file_id = '1NM1UDdZnwHwiNDxhcP-nndaWj24m-90L'
destination = './m2det512_vgg.pth'
download_file_from_google_drive(file_id, destination)

3. Copy image file

--Mount Google Drive --After that, copy all the jpg files in the folder containing the image files to "imgs" that executes M2Det (in my environment, the image files are stored under My Drive> ML> work).

Google Drive mount

from google.colab import drive
drive.mount('/content/drive')

Copy image file

!cp /content/drive/My\ Drive/ML/work/*.jpg ./imgs

4. Run the model

--Run the model

Model execution

!python demo.py -c=configs/m2det512_vgg.py -m=m2det512_vgg.pth

5. Display results

--Display the result of execution --The executed image file will be created as "XXX_m2det.jpg "

python

import cv2
import matplotlib.pyplot as plt

plt.figure(figsize=(5, 5), dpi=200)
img = cv2.imread('imgs/herd_of_horses_m2det.jpg')
show_img = cv2.cvtColor(img, cv2.COLOR_BGR2RGB)
plt.imshow(show_img)

--Other images were also carried out

Source code

https://github.com/hiraku00/m2det_test

reference

-I tried M2Det, which is rumored to be faster than YOLO and SSD -Try demo / evaluation / learning with M2Det -I tried to move the latest and strongest object detection technology M2Det