π
Banana Recognizer Processing power used:
- GPU: 2x NVIDIA Tesla V100 32GB
- CPU: 2x Intel Xeon Gold 6132 2.6G, 14C/28T, 10.4GT/s 2UPI, 19M Cache, Turbo, HT (140W) DDR4-2666
- RAM: 24x 32GB = 768GB
Convolutional Neural Network: `BananaNetwork/banana_net_17_12_18.h5
INPUT > CONV > RELU > POOL > FC > RELU > FC
- Build script: `CNN.py
Convolutional Netural Network: BananaNetwork/le_banana_net_26_02_19.h5
-
INPUT > CONV > RELU > POOL > CONV > RELU > POOL > FC > RELU > FC
-
Build script:
CNN_LeNet.py
-
Also looking at other alternatives for design of network
File structure
βββ BananaNetwork
β βββ banana_net.h5
β βββ banana_net_17_12_18.h5
β βββ banana_net_26_02_19.h5
β βββ le_banana_net.h5
βββ CNN.py
βββ CNN_LeNet.py
βββ README.md
βββ WebcamApp
β βββ camera_app.py
βββ banana_predicter.py
βββ datasets
β βββ google_images
β β βββ ...
β βββ google_synset_fruit365
β β βββ ...
β βββ synset_fruit365
β βββ ...
βββ epoch_fig.png
βββ google_images_scraper.js
βββ gui
β βββ gui.py
β βββ gui.ui
βββ image_downloader.py
βββ network_trainer.py
βββ predict_tests
β βββ ...
βββ urls
βββ google_banana_urls.txt
βββ synset_banana_urls.txt
Datasets
- Fruit365
- Synset - ImageNet
- Google Images