Tobenna-KA/mimeai-api

Install Dependencies

pip3 install -r requirements.txt

Start App

• export FLASK_APP=wisgi.py
• export FLASK_ENV=development
• flask run

Docker

• docker rm -f $(docker ps -a -q)

client side can be found at link

Sample report

---

Layer (type)          Output Shape          Param #          Connected to

================================================================================================== input_1          (InputLayer)          (None, 28, 28, 3)          0

---

conv1 (Conv2D)          (None, 20, 20, 128)          31232          input_1[0][0]

---

primarycap_conv2d          (Conv2D)          (None, 6, 6, 256)          2654464          conv1[0][0]

---

primarycap_reshape          (Reshape)          (None, 1152, 8)          0          primarycap_conv2d[0][0]

---

primarycap_squash          (Lambda)          (None, 1152, 8)          0          primarycap_reshape[0][0]

---

digitcaps (CapsuleLayer)          (None, 10, 16)          1474560          primarycap_squash[0][0]

---

mask_2          (Mask)          (None, 160)          0          digitcaps[0][0]

---

capsnet          (Length)          (None, 10)          0          digitcaps[0][0]

---

decoder          (Sequential)          (None, 28, 28, 3)          3018544          mask_2[0][0]

============================================================================================= Total params: 7,178,800 Trainable params: 7,178,800 Non-trainable params: 0

sample response

    {
        "disease": "Early_blight",
        "accuracy": 92, 
        "image_path": "static/predicted_images/Early_blight/2020-06-11 23:19:46.142173.png",
        "success": true
    }