Returns fruit type given an image with about 85% accuracy.
See https://www.kaggle.com/code/dima806/fruit-100-types-image-detection-vit for more details.
Classification report:
precision recall f1-score support
abiu 0.7799 0.9056 0.8380 180
acai 0.8118 0.8389 0.8251 180
acerola 0.8701 0.8556 0.8627 180
ackee 0.9451 0.9556 0.9503 180
ambarella 0.5696 0.7278 0.6390 180
apple 0.9027 0.9278 0.9151 180
apricot 0.7046 0.9278 0.8010 180
avocado 0.9297 0.9556 0.9425 180
banana 0.9781 0.9944 0.9862 180
barbadine 0.9074 0.5444 0.6806 180
barberry 0.8122 0.8889 0.8488 180
betel_nut 0.9420 0.7222 0.8176 180
bitter_gourd 0.9888 0.9833 0.9861 180
black_berry 0.5260 0.9000 0.6639 180
black_mullberry 0.9641 0.8944 0.9280 180
brazil_nut 0.9298 0.8833 0.9060 180
camu_camu 0.8325 0.9111 0.8700 180
cashew 0.9889 0.9889 0.9889 180
cempedak 0.9706 0.5500 0.7021 180
chenet 0.7422 0.9278 0.8247 180
cherimoya 0.5869 0.6944 0.6361 180
chico 0.5940 0.4389 0.5048 180
chokeberry 0.8444 0.8444 0.8444 180
cluster_fig 0.9236 0.8056 0.8605 180
coconut 0.9167 0.9778 0.9462 180
corn_kernel 0.9781 0.9944 0.9862 180
cranberry 0.9067 0.7556 0.8242 180
cupuacu 0.8846 0.8944 0.8895 180
custard_apple 0.5000 0.0056 0.0110 180
damson 0.8687 0.9556 0.9101 180
dewberry 0.7869 0.2667 0.3983 180
dragonfruit 0.9890 0.9944 0.9917 180
durian 0.9730 1.0000 0.9863 180
eggplant 0.9833 0.9833 0.9833 180
elderberry 0.9553 0.9500 0.9526 180
emblic 0.8927 0.8778 0.8852 180
feijoa 0.9111 0.9111 0.9111 180
fig 0.8696 1.0000 0.9302 180
finger_lime 0.9647 0.9111 0.9371 180
gooseberry 0.8966 0.8667 0.8814 180
goumi 0.8020 0.9000 0.8482 180
grape 0.9661 0.9500 0.9580 180
grapefruit 0.8696 0.7778 0.8211 180
greengage 0.8434 0.7778 0.8092 180
grenadilla 0.6457 0.8000 0.7146 180
guava 0.8122 0.8889 0.8488 180
hard_kiwi 0.8367 0.9111 0.8723 180
hawthorn 0.8246 0.7833 0.8034 180
hog_plum 0.8667 0.0722 0.1333 180
horned_melon 0.9943 0.9722 0.9831 180
indian_strawberry 0.5427 0.4944 0.5174 180
jaboticaba 0.9480 0.9111 0.9292 180
jackfruit 0.6917 0.9722 0.8083 180
jalapeno 0.9728 0.9944 0.9835 180
jamaica_cherry 0.9136 0.8222 0.8655 180
jambul 0.8750 0.8556 0.8652 180
jocote 0.7365 0.6056 0.6646 180
jujube 0.8554 0.7889 0.8208 180
kaffir_lime 0.9672 0.9833 0.9752 180
kumquat 0.8000 0.9333 0.8615 180
lablab 0.9835 0.9944 0.9890 180
langsat 0.8656 0.8944 0.8798 180
longan 0.9016 0.9667 0.9330 180
mabolo 0.9405 0.8778 0.9080 180
malay_apple 0.6173 0.5556 0.5848 180
mandarine 0.7811 0.8722 0.8241 180
mango 0.8071 0.8833 0.8435 180
mangosteen 0.9609 0.9556 0.9582 180
medlar 0.9503 0.9556 0.9529 180
mock_strawberry 0.5568 0.5722 0.5644 180
morinda 0.9727 0.9889 0.9807 180
mountain_soursop 0.9496 0.7333 0.8276 180
oil_palm 0.9053 0.9556 0.9297 180
olive 0.9704 0.9111 0.9398 180
otaheite_apple 0.5736 0.6278 0.5995 180
papaya 0.7882 0.8889 0.8355 180
passion_fruit 0.7720 0.8278 0.7989 180
pawpaw 0.8428 0.7444 0.7906 180
pea 0.9375 1.0000 0.9677 180
pineapple 1.0000 1.0000 1.0000 180
plumcot 0.8525 0.5778 0.6887 180
pomegranate 0.9418 0.9889 0.9648 180
prikly_pear 0.9834 0.9889 0.9861 180
quince 0.9399 0.9556 0.9477 180
rambutan 1.0000 1.0000 1.0000 180
raspberry 0.9206 0.9667 0.9431 180
redcurrant 0.9040 0.9944 0.9471 180
rose_hip 0.8595 0.8833 0.8712 180
rose_leaf_bramble 0.9050 0.9000 0.9025 180
salak 0.8947 0.9444 0.9189 180
santol 0.8870 0.8722 0.8796 180
sapodilla 0.5727 0.7222 0.6388 180
sea_buckthorn 0.9780 0.9889 0.9834 180
strawberry_guava 0.8407 0.8500 0.8453 180
sugar_apple 0.4711 0.9500 0.6298 180
taxus_baccata 0.9676 0.9944 0.9808 180
ugli_fruit 0.9202 0.8333 0.8746 180
white_currant 1.0000 1.0000 1.0000 180
yali_pear 0.9448 0.9500 0.9474 180
yellow_plum 0.7552 0.8056 0.7796 180
accuracy 0.8498 18000
macro avg 0.8570 0.8498 0.8417 18000
weighted avg 0.8570 0.8498 0.8417 18000
- Downloads last month
- 9
This model does not have enough activity to be deployed to Inference API (serverless) yet. Increase its social
visibility and check back later, or deploy to Inference Endpoints (dedicated)
instead.
Model tree for dima806/fruit_100_types_image_detection
Base model
google/vit-base-patch16-224-in21k