Show and Tell: A Neural Image Caption Generator - Paper Replication

Vinyals, Oriol, et al. "Show and tell: A neural image caption generator." Proceedings of the IEEE conference on computer vision and pattern recognition. 2015.

Results

Image	Captions
Example 1	Actual Caption: <start> <oov> <oov> to <oov> the image above <end> <oov> Output Caption: <start> cat laying on top of laptop <end>
Example 2	Actual Caption: <start> <oov> <oov> slope near <oov> <oov> <end> <oov> <oov> Output Caption: <start> person that is skiing in the snow <end>
Example 3	Actual Caption: <start> bus is parked at the bus <oov> <end> <oov> Output Caption: <start> large red bus on city street <end>

---

Metrics

The performance of the model was evaluated using standard metric for image captioning tasks, the BLEU scores. Below are the results:

• BLEU-1: 44
• BLEU-2: 22
• BLEU-4: 9

These metrics indicate the effectiveness of the caption generator in producing relevant and accurate captions.

License

Distributed under the MIT License. See LICENSE.txt for more information.

(back to top)

Contact

(back to top)