zeromodels/inception_v3_tf_in1k

See our collection for all versions of Inception-V3.

Run Inception-V3 with Keras 3: JAX, PyTorch, or TensorFlow

kerasformers/inception_v3_tf_in1k

Paper: Rethinking the Inception Architecture for Computer Vision (arXiv:1512.00567) · HF Papers

Inception-V3 factorizes convolutions for efficient multi-scale features. Classifier or multi-stage backbone.

For more details on the model, please go to the upstream model card.

Pure-Keras 3 conversion of timm/inception_v3.tf_in1k for kerasformers. One implementation runs unmodified on TensorFlow / Torch / JAX.

This is an image-classification / backbone checkpoint (InceptionV3ImageClassify / InceptionV3Model).

✨ Quick start

import os
os.environ["KERAS_BACKEND"] = "torch"  # or "jax" / "tensorflow"

from PIL import Image
import numpy as np
from kerasformers.models.inceptionv3 import InceptionV3ImageClassify, InceptionV3Model

model = InceptionV3ImageClassify.from_weights("kerasformers/inception_v3_tf_in1k")
backbone = InceptionV3Model.from_weights(
    "kerasformers/inception_v3_tf_in1k", as_backbone=True
)

image = Image.open("your_image.jpg").convert("RGB")
image = image.resize((299, 299))
x = np.asarray(image, dtype="float32")[None]  # (1, H, W, 3)
print(model(x).shape)  # (1, num_classes)
feats = backbone(x)
print(len(feats), [tuple(f.shape) for f in feats])

Load any Inception-V3 variant the same way with from_weights("kerasformers/<variant>"):

Variant	Hub
inception_v3_gluon_in1k	kerasformers/inception_v3_gluon_in1k
inception_v3_tf_adv_in1k	kerasformers/inception_v3_tf_adv_in1k
inception_v3_tf_in1k	kerasformers/inception_v3_tf_in1k

Tips

• Set KERAS_BACKEND before importing Keras / kerasformers.
• InceptionV3ImageClassify returns class logits; InceptionV3Model returns features (as_backbone=True for multi-scale stages).
• See docs and Loading Weights.
• Upstream / timm checkpoints: InceptionV3ImageClassify.from_weights("hf:timm/inception_v3.tf_in1k").

Special Thanks

A huge thank you to the Inception-V3 authors and the timm / Hub communities for creating and releasing these models.

License: see YAML license (usually matches the upstream checkpoint).