Services ======== This section describes solutions offered by instyle.ai and their use cases. For details on integration with our services please refer to :ref:`integration` section. .. _similarity: Similarity ---------- Recommend visually similar alternatives. All fashion product categories are supported (incl. accessories). |similarity_example| Use case: product detail page recommender ######################################### .. |similarity_example| image:: _static/similarity-example.png :width: 800px Integration notes ################# - :ref:`pdp_recoms` - :ref:`catalog_v2` (batch only) .. _personalized_recoms: Personalized recommendations ---------------------------- Recommend the most relevant products for a given user based on his viewing and purchase history. All fashion product categories are supported (incl. accessories). Personalized similarity ####################### |personalized_example_jpg| .. |personalized_example_jpg| image:: _static/personalized-example.jpg :width: 800px Last seen ######### Show the last visited products for user convenience. Integration notes ################# - :ref:`pdp_recoms` .. _shop_the_look: Shop the look ------------- Recommend complimentary products to form a fashionable outfit. |shop_the_look_example| .. |shop_the_look_example| image:: _static/shop-the-look-example.png :width: 800px Integration notes ################# - :ref:`pdp_recoms` - :ref:`catalog_v2` (batch only) .. _tags: Product Tags ------------ Extract visual attributes like color or sleeve length from product images. Supported product categories: top, skirt, dress, sweater, hoodie, jacket, pants, shorts, shoe, handbag, bikinis top, bikinis bottom, swimsuit. |tags_example| .. in order to update tags docs download the client tags dictionary from google docs and run download docx from https://docs.google.com/document/d/1UUGt8YSKLgyXfr_pDvc7V1eKzPc75FNZ9pQmihYyZww/edit# cd docs/source pandoc "CLIENT Tag labels dictionary.docx" -f docx -t rst -o tags_dict.rst --extract-media ./_static/product_tags --columns 300 **Check out details on available product tags** .. toctree:: :maxdepth: 1 :glob: tags_dict Use case: product filtering ########################### Use case: tags for full-text search ################################### .. |tags_example| image:: _static/tags_example.jpg :width: 375px Integration notes ################# - :ref:`catalog_v2` (batch only) .. _content_tags: Content Tags ------------ Extract visual information about the product photo composition. All fashion product categories are supported, except for accessories. |content_tags_example| **Check out details on available content tags** .. toctree:: :maxdepth: 1 :glob: content_tags_dict .. download docx content tags dict from https://docs.google.com/document/d/16_NUfpQNYrXhzG2_WBpxHnm73wf09YxVloVCReocCzQ/edit?usp=sharing pandoc "CLIENT_content_tags.docx" -f docx -t rst -o content_tags_dict.rst --extract-media ./_static/content_tags --columns 300 .. todo: add image for content_tag-example.png can we add image flipping animation? Use case: On hover animation ############################ Automatically pick image pairs for on-hover animations .. raw:: html
Rotate
Zoom in
Zoom out
Use case: Primary image selection ################################# Automatically determine the best primary image. .. raw:: html
.. raw:: html
.. |content_tags_example| image:: _static/content_tags_example_2.jpg :width: 400px Integration notes ################# - :ref:`catalog_v2` (batch only) .. _image_normalization: Image Normalization ------------------- Curate product images from multiple sources to achieve smooth and consistent listing page. We will select the best images and unify their backgrounds, product positions and shadows. All fashion product categories are supported, except for accessories. Example tops ############ |rembg_example_1| Example pants ############# |rembg_example_2| Example shoes ############# |rembg_example_3| Integration notes ################# - :ref:`catalog_v2` (batch only) - field `curated_images` under `look` entity .. |rembg_example_1| image:: _static/rembg_example_1.jpg :width: 800px .. |rembg_example_2| image:: _static/rembg_example_2.jpg :width: 800px .. |rembg_example_3| image:: _static/rembg_example_3.jpg :width: 800px .. _street2shop: Street2Shop ----------- Search for products by image. Supported product categories: top, skirt, dress, sweater, hoodie, jacket, pants, shorts, shoe, handbag, bikinis top, bikinis bottom, swimsuit. |s2s_example| Use case: Search by Image ######################### With our search by image technology, shoppers can simply snap a picture or upload one from their device, or even use images from social media posts they liked, and our AI will match it with products on your website, making it easier for them to find what they are looking for. Use case: Recommend complementary products ########################################## Frequently, a product photo features multiple products that are well matched with the main garment. Most likely, the customer already likes the whole look and using our search by image technology on product images, we can automatically recommend the rest of the outfit. Integration notes ################# - :ref:`street2shop_api` .. |s2s_example| image:: _static/s2s_example.jpg :width: 800px Category classification ----------------------- Utilizing advanced algorithms, our service automatically categorizes products based on attributes, descriptions, and images. Say goodbye to manual sorting and enjoy efficient, accurate categorization that enhances user experience. With customizable settings and seamless scalability, your online store will present a consistent, organized selection that keeps customers engaged and satisfied. Integration notes ################# - :ref:`category_classification` .. _ads_optimization: Google Ads optimization ----------------------- Extracted visual attributes (tags) can be also beneficial for improving ads efficiency. We can match the detected tags with frequent search terms and compose unique titles with enhanced ads performance. For the maximum efficiency we will generate multiple titles per product and spread them across size variants to cover the whole search space and match the search term distribution. |ads_optimization_example| .. |ads_optimization_example| image:: _static/ads_optimization_example.png :width: 800px