This resource contains a segmentation model for the identification of polyps during colonoscopies, as well as a sample colonoscopy video for inference and testing.
This model was trained on the Kvasir-SEG dataset , using the ColonSegNet model architecture .
 Jha, Debesh, Pia H. Smedsrud, Michael A. Riegler, Pål Halvorsen, Thomas de Lange, Dag Johansen, and Håvard D. Johansen, "Kvasir-seg: A segmented polyp dataset" Proceedings of the International Conference on Multimedia Modeling, pp. 451-462, 2020.
 Jha D, Ali S, Tomar NK, Johansen HD, Johansen D, Rittscher J, Riegler MA, Halvorsen P. Real-Time Polyp Detection, Localization and Segmentation in Colonoscopy Using Deep Learning. IEEE Access. 2021 Mar 4;9:40496-40510. doi: 10.1109/ACCESS.2021.3063716. PMID: 33747684; PMCID: PMC7968127.
The colonoscopy video sample sequence was provided by Simula Research Laboratory. The video sample includes a scene with a polyp identified by the model.
The AI model for polyp identification is a segmentation model which, given an RGB image of
512 x 512, provides semantic segmentation of the polyps.
Each pixel stores a confidence score of [0,1] for polyp presence.
For details on the model inputs and outputs see below.
INPUT__0- Input RGB image (batchsize, height, width, channels)
shape=[1, 512, 512, 3]
output_old- Segmentation output with per-pixel confidence [0,1].
shape=[1, 512, 512]
The package contains two folders:
modelfolder with the source model in ONNX format, and .engine files for the TensorRT models optimized for the Holoscan Developer Kits.
videofolder with the recorded video data, both in
mp4(source) and converted to the NVIDIA GXF tensor format.
/ ├── NVIDIA-Clara-Holoscan-SDK-EULA.pdf ├── NVIDIA Colonoscopy Software License (9Sept2022).docx ├── model │ ├── colon.onnx │ └── engine │ ├── NVIDIA-RTX-A6000_c86_n84.engine │ └── Quadro-RTX-6000_c75_n72.engine └── video ├── colon_exam_720x576.mp4 ├── colonoscopy.gxf_entities └── colonoscopy.gxf_index
Refer to the license agreement for use of the sample data.