CountCLIP -- [Re] Teaching CLIP to Count to Ten
This is an incremental reproducibility study addressing the lack of quantitative understanding in vision-language models for researchers in computer vision.
The paper reproduces a method to finetune CLIP for improved zero-shot counting accuracy in images while maintaining classification performance, achieving enhanced results with a smaller training subset and lower computational resources.
Large vision-language models (VLMs) are shown to learn rich joint image-text representations enabling high performances in relevant downstream tasks. However, they fail to showcase their quantitative understanding of objects, and they lack good counting-aware representation. This paper conducts a reproducibility study of 'Teaching CLIP to Count to Ten' (Paiss et al., 2023), which presents a method to finetune a CLIP model (Radford et al., 2021) to improve zero-shot counting accuracy in an image while maintaining the performance for zero-shot classification by introducing a counting-contrastive loss term. We improve the model's performance on a smaller subset of their training data with lower computational resources. We verify these claims by reproducing their study with our own code. The implementation can be found at https://github.com/SforAiDl/CountCLIP.